d29c2c2fb7
Model: mlfoundations-dev/mlfoundations-dev_code-stratos-unverified-scaled-0_25_stratos_7b Source: Original Platform
39 lines
1.1 KiB
YAML
39 lines
1.1 KiB
YAML
assistant_tag: assistant
|
|
bf16: true
|
|
content_tag: value
|
|
cutoff_len: 16384
|
|
dataset: /p/data1/mmlaion/marianna/lf_datasets/mlfoundations-dev/code-stratos-unverified-scaled-0.25
|
|
dataset_dir: ONLINE
|
|
ddp_timeout: 180000000
|
|
deepspeed: dcft/train/zero3_offload.json
|
|
do_train: true
|
|
enable_liger_kernel: true
|
|
finetuning_type: full
|
|
formatting: sharegpt
|
|
global_batch_size: 96
|
|
gradient_accumulation_steps: 2
|
|
hub_model_id: mlfoundations-dev/mlfoundations-dev_code-stratos-unverified-scaled-0_25_stratos_7b
|
|
learning_rate: 1.0e-05
|
|
logging_steps: 1
|
|
lr_scheduler_type: cosine
|
|
max_samples: 1000000
|
|
messages: conversations
|
|
model_name_or_path: Qwen/Qwen2.5-7B-Instruct
|
|
num_train_epochs: 3.0
|
|
output_dir: /p/data1/mmlaion/marianna/dcft_checkpoints/mlfoundations-dev_code-stratos-unverified-scaled-0.25_stratos_7b
|
|
overwrite_cache: true
|
|
per_device_train_batch_size: 1
|
|
plot_loss: true
|
|
preprocessing_num_workers: 16
|
|
push_to_db: false
|
|
push_to_hub: false
|
|
report_to: wandb
|
|
role_tag: from
|
|
run_name: Bespoke-Stratos-17k
|
|
save_steps: 100
|
|
stage: sft
|
|
template: qwen25
|
|
use_unsloth_gc: true
|
|
user_tag: user
|
|
warmup_ratio: 0.1
|