q3-8b-train_final_v2_nb2_mt8192_replaced_fix/slurm/9168643.0.err

./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
./slurm_setup.sh: line 48: GLOBAL_TRAIN_BATCH_SIZE / NNODES / GPUS_PER_NODE / GRADIENT_ACCUMULATION_STEPS: division by 0 (error token is "GRADIENT_ACCUMULATION_STEPS")
2026-04-13 23:07:49,112 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,112 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,131 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,131 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,131 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,131 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,135 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,135 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,144 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,144 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,144 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,144 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,153 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,153 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,164 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,164 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,165 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,165 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,165 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,165 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,165 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,165 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,170 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,170 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,171 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,171 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,172 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,172 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,175 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,175 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,180 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,180 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,190 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,190 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,192 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,192 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,192 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,193 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,197 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,197 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,201 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,201 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,201 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,202 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,203 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,203 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,204 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,205 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,211 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,211 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,219 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,219 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,219 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,219 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,222 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,222 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,222 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,223 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,228 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,228 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,228 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,228 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,230 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,230 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,239 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,239 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,246 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,246 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,254 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,254 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,260 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,260 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,268 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,268 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,274 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,274 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,277 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,277 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,277 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,277 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,278 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,278 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,281 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,281 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,283 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,283 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,295 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,295 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,312 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,313 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,313 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,313 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,313 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,313 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,313 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,313 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,313 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,314 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,323 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,323 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,323 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,323 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,325 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,325 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,334 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,334 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,334 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,334 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,338 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,338 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:49,338 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:49,338 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:49,443 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,459 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,459 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,465 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,476 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,477 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,477 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,483 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,483 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,489 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,489 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,489 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,497 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,497 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,498 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,505 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,516 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,517 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,523 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,525 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,529 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,530 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,535 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,536 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,534 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,537 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,537 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,542 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,543 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,551 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,551 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,553 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,561 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,581 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,588 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,588 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,599 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,600 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,602 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,602 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,607 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,611 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,611 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,628 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,628 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,630 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,637 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,637 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,642 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,648 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,648 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,649 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,652 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,652 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,656 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:49,656 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:52,411 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,411 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,411 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,411 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,411 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,411 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:52,411 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:52,411 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,411 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,411 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:52,411 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:52,411 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:52,411 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:52,411 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
2026-04-13 23:07:52,412 - INFO - Note: detected 255 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable.
2026-04-13 23:07:52,412 - INFO - Note: NumExpr detected 255 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
/home/ligengz/anaconda3/lib/python3.11/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.5' currently installed).
  from pandas.core import (
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:07:53,556 - INFO - PyTorch version 2.6.0 available.
2026-04-13 23:08:06,156 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,165 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,200 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,219 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,277 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,277 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,281 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,282 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,289 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,291 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,295 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,298 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,302 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,302 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,305 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,307 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,312 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,299 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,327 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,339 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,339 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,360 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,365 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,387 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,408 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,462 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,486 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,484 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,486 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,504 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,506 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,507 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:06,526 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-06_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:12,950 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:12,980 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:12,982 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:12,982 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,014 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,034 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,036 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,037 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,038 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,038 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3848', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,042 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,052 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,059 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,083 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3227', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,246 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:13,249 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:17,274 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1015', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:17,329 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3297', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:17,338 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:17,338 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1027', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:17,353 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-1134', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:17,434 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:17,434 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-05_batch-block1-3273', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,261 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,387 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 2, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,388 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 5, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,390 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 4, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,391 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 1, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,391 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 3, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,392 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 7, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules
2026-04-13 23:08:32,393 - INFO - Training config: {'model_name': 'runs/dev/tw-8node-resume/checkpoint-64', 'template_name': 'qwen', 'block_size': 40960, 'wandb_project': 'ThreadWeaver', 'train_file_path': './data-04-13', 'dagger': False, 'attn_implementation': 'flex_attention', 'output_dir': 'runs/dev/tw-8node-resume', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': False, 'do_predict': False, 'eval_strategy': <IntervalStrategy.NO: 'no'>, 'prediction_loss_only': False, 'per_device_train_batch_size': 1, 'per_device_eval_batch_size': 1, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0001, 'adam_beta1': 0.9, 'adam_beta2': 0.95, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 8.0, 'max_steps': -1, 'lr_scheduler_type': <SchedulerType.COSINE: 'cosine'>, 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.05, 'warmup_steps': 0, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': 'runs/dev/tw-8node-resume/runs/Apr13_23-08-30_batch-block1-3887', 'logging_strategy': <IntervalStrategy.STEPS: 'steps'>, 'logging_first_step': False, 'logging_steps': 1.0, 'logging_nan_inf_filter': True, 'save_strategy': <SaveStrategy.STEPS: 'steps'>, 'save_steps': 20, 'save_total_limit': 2, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': None, 'local_rank': 6, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': None, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': 'runs/dev/tw-8node-resume', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'tp_size': 0, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, 'deepspeed': 'configs/deepspeed_zero3.json', 'label_smoothing_factor': 0.0, 'optim': <OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, 'optim_args': None, 'adafactor': False, 'group_by_length': False, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': 'runs/dev/tw-8node-resume/checkpoint-64', 'hub_model_id': None, 'hub_strategy': <HubStrategy.EVERY_SAVE: 'every_save'>, 'hub_token': None, 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': True, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': None, '_n_gpu': 1, 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules

Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.62s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  2
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  2
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.59s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.60s/it]
Loading checkpoint shards:  2
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.62s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.62s/it]
Loading checkpoint shards:  25%|██▌       | 1/4 [00:01<00:04,  1.61s/it]
Loading checkpoint shards:  
▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:05<00:01,  1.68s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
▌  | 3/4 [00:04<00:01,  1.64s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.64s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.64s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.64s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.64s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.64s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.23s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
▌  | 3/4 [00:04<00:01,  1.66s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.66s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.66s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.66s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.66s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.66s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards:  75%|███████▌  | 3/4 [00:04<00:01,  1.67s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]



Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.26s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.23s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.23s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.23s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.23s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.23s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.23s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.26s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]



Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]


Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.25s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:04<00:00,  1.24s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.27s/it]

Loading checkpoint shards:  75%|███████▌  | 3/4 [00:05<00:01,  1.72s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.22s/it]
Loading checkpoint shards: 100%|██████████| 4/4 [00:05<00:00,  1.42s/it]
2026-04-13 23:08:44,037 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,045 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,106 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,107 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,108 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,107 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,108 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,108 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,108 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,108 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,109 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,109 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,109 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,109 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,110 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,112 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,111 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,113 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,113 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,113 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,113 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,113 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,113 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,114 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,114 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,114 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,115 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,116 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,115 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,115 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,115 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,115 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,116 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,116 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,115 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,116 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,115 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,116 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,116 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,116 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,116 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,117 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,117 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,117 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,117 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,119 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,119 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,119 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,121 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,120 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,120 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,121 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,121 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,121 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,121 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,121 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,122 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,123 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,122 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,122 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,122 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,122 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,122 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,123 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,123 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,123 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,123 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,123 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,123 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,125 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,124 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,125 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,125 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,125 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,125 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,126 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,126 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,126 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,126 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,126 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,126 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,127 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,128 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,127 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,128 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,129 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,130 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,130 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,129 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,129 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,130 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,131 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,131 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,131 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,131 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,132 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,131 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,132 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,132 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,132 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,133 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,133 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,134 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,133 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,134 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,135 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,134 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,136 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,137 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,138 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,138 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,139 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,270 - INFO - Applying Liger kernels to model instance with model type: qwen3 with kwargs: {}
2026-04-13 23:08:44,281 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,284 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,284 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,284 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,284 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,284 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,284 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
2026-04-13 23:08:44,285 - INFO - Resuming trainer state from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
wandb: WARNING The `run_name` is currently set to the same value as `TrainingArguments.output_dir`. If this was not intended, please specify a different run name by setting the `TrainingArguments.run_name` parameter.
wandb: Currently logged in as: ligeng-zhu to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
wandb: Tracking run with wandb version 0.21.0
wandb: Run data is saved locally in runs/dev/tw-8node-resume/wandb/run-20260413_230852-tw-8node-resume
wandb: Run `wandb offline` to turn off syncing.
wandb: Resuming run runs/dev/tw-8node-resume
wandb: ⭐️ View project at https://wandb.ai/ligeng-zhu/ThreadWeaver
wandb: 🚀 View run at https://wandb.ai/ligeng-zhu/ThreadWeaver/runs/tw-8node-resume
A ConfigError was raised whilst setting the number of model parameters in Weights & Biases config.

  0%|          | 0/64 [00:00<?, ?it/s]
                                      

  0%|          | 0/64 [00:00<?, ?it/s]
  0%|          | 0/64 [00:00<?, ?it/s]