1631 lines
268 KiB
Plaintext
1631 lines
268 KiB
Plaintext
|
|
|||
|
|
The following have been reloaded with a version change:
|
|||
|
|
1) GCCcore/.14.3.0 => GCCcore/14.3.0
|
|||
|
|
|
|||
|
|
|
|||
|
|
Lmod is automatically replacing "GCC/14.3.0" with
|
|||
|
|
"nvidia-compilers/25.9-CUDA-13".
|
|||
|
|
|
|||
|
|
Deactivating conda environment: /e/scratch/jureap59/feuer1/miniforge3/envs/otagent
|
|||
|
|
Activating RL environment: /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl
|
|||
|
|
Python executable: /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/bin/python
|
|||
|
|
Python path check: /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/bin/python
|
|||
|
|
[ray] RAY_TMPDIR=/tmp/ray/ray_485102
|
|||
|
|
[triton_cache] Triton cache: /tmp/triton_cache_feuer1_485102
|
|||
|
|
[triton_cache] TorchInductor cache: /tmp/torchinductor_cache_feuer1_485102
|
|||
|
|
[proxy] ✓ Found proxychains binary at /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4
|
|||
|
|
[proxy] Setting up SSH tunnel to jpbl-s01-01
|
|||
|
|
[proxy] SSH key: /e/home/jusers/feuer1/jupiter/.ssh/authorized_keys/id_ed25519_jsc
|
|||
|
|
[proxy] Tunnel port: 7003
|
|||
|
|
[proxy] Node IP: 10.128.43.81 (workers will connect here)
|
|||
|
|
[proxy] ✓ SSH tunnel started successfully
|
|||
|
|
[proxy] ✓ Generated proxychains config at /e/home/jusers/feuer1/jupiter/.proxychains/proxychains_485102.conf
|
|||
|
|
[proxy] - Internal traffic (10.x.x.x, 172.x.x.x, 169.254.x.x) → DIRECT
|
|||
|
|
[proxy] - External traffic (internet) → PROXY via tunnel
|
|||
|
|
[proxy] ✓ Daytona timeout settings configured
|
|||
|
|
[proxy] Testing proxy connectivity...
|
|||
|
|
[proxychains] config file found: /e/home/jusers/feuer1/jupiter/.proxychains/proxychains_485102.conf
|
|||
|
|
[proxychains] preloading /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/lib/libproxychains4.so
|
|||
|
|
[proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e
|
|||
|
|
[proxy] ✓ Proxy connectivity test passed (huggingface.co reachable via wrapped binary)
|
|||
|
|
[proxy] ⚠ Tunnel not accessible at 10.128.43.81:7003 (workers may fail)
|
|||
|
|
[proxy] ✓ Proxy setup complete (using wrapped binary for Ray workers)
|
|||
|
|
[container_runtime] Using cloud backend: daytona (no local container setup)
|
|||
|
|
=== Universal RL Training Runner ===
|
|||
|
|
Config: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/configs/a3-rl-DCAgent_code-contests-noblock_rl_config.json
|
|||
|
|
Working directory: /e/scratch/jureap59/feuer1/OpenThoughts-Agent
|
|||
|
|
Python: /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/bin/python
|
|||
|
|
Python version: Python 3.12.12
|
|||
|
|
Proxy: DISABLED (direct internet or not configured)
|
|||
|
|
========================================
|
|||
|
|
=== RLJobRunner: a3-rl-DCAgent_code-contests-noblock ===
|
|||
|
|
[wandb_utils] Fixing permissions on: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/wandb
|
|||
|
|
[wandb_utils] WandB directory ready: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/wandb
|
|||
|
|
HF_TOKEN=****pDbg
|
|||
|
|
HF_HUB_CACHE=/e/data1/datasets/playground/ot-baf/hf_hub
|
|||
|
|
SUPABASE_URL=https://rpzmyuapoqilpghynmza.s... (direct Supabase config)
|
|||
|
|
Environment configured:
|
|||
|
|
TENSOR_PARALLEL_SIZE=1
|
|||
|
|
NUM_INFERENCE_ENGINES=56
|
|||
|
|
POLICY_NUM_NODES=14
|
|||
|
|
WANDB_DIR=/e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/wandb
|
|||
|
|
Starting Ray cluster with 14 nodes, 4 GPUs/node
|
|||
|
|
Cleaning up existing Ray instances...
|
|||
|
|
=== Starting Ray Cluster ===
|
|||
|
|
Nodes: 14
|
|||
|
|
GPUs per node: 4
|
|||
|
|
CPUs per node: 288
|
|||
|
|
Head node: jpbo-078-33 (10.128.43.81)
|
|||
|
|
Ray port: 6379
|
|||
|
|
============================
|
|||
|
|
Starting Ray head on jpbo-078-33 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_head_jpbo-078-33.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.81 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-33 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --head --node-ip-address=10.128.43.81 --port=6379 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray head on jpbo-078-33
|
|||
|
|
Starting Ray worker on jpbo-078-36 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-36.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.84 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-36 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.84 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 1 on jpbo-078-36
|
|||
|
|
Starting Ray worker on jpbo-078-37 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-37.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.85 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-37 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.85 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 2 on jpbo-078-37
|
|||
|
|
Starting Ray worker on jpbo-078-38 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-38.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.86 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-38 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.86 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 3 on jpbo-078-38
|
|||
|
|
Starting Ray worker on jpbo-078-39 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-39.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.87 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-39 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.87 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 4 on jpbo-078-39
|
|||
|
|
Starting Ray worker on jpbo-078-40 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-40.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.88 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-40 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.88 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 5 on jpbo-078-40
|
|||
|
|
Starting Ray worker on jpbo-078-42 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-42.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.90 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-42 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.90 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 6 on jpbo-078-42
|
|||
|
|
Starting Ray worker on jpbo-078-43 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-43.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.91 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-43 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.91 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 7 on jpbo-078-43
|
|||
|
|
Starting Ray worker on jpbo-078-45 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-45.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.93 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-45 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.93 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 8 on jpbo-078-45
|
|||
|
|
Starting Ray worker on jpbo-078-46 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-46.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.94 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-46 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.94 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 9 on jpbo-078-46
|
|||
|
|
Starting Ray worker on jpbo-078-47 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-47.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.95 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-47 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.95 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 10 on jpbo-078-47
|
|||
|
|
Starting Ray worker on jpbo-078-48 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-078-48.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.96 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-078-48 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.96 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 11 on jpbo-078-48
|
|||
|
|
Starting Ray worker on jpbo-079-07 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-079-07.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.103 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-079-07 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.103 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 12 on jpbo-079-07
|
|||
|
|
Starting Ray worker on jpbo-079-12 (logging to /e/scratch/jureap59/feuer1/OpenThoughts-Agent/experiments/logs/ray_worker_jpbo-079-12.log)...
|
|||
|
|
Command: srun --export=ALL,VLLM_HOST_IP=10.128.43.108 --nodes=1 --ntasks=1 --gres=gpu:4 --gpu-bind=none --overlap --cpu-bind=none -w jpbo-079-12 bash -c /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f "$PROXYCHAINS_CONF_FILE" ray start --address=10.128.43.81:6379 --node-ip-address=10.128.43.108 --num-gpus=4 --num-cpus=288 --block --object-store-memory=42949672960
|
|||
|
|
Started Ray worker 13 on jpbo-079-12
|
|||
|
|
Waiting for cluster (56 GPUs, 14 nodes)...
|
|||
|
|
Connecting to Ray at 10.128.43.81:6379 (expecting 14 nodes, 56.0 GPUs)
|
|||
|
|
Ray connection established, polling for resources...
|
|||
|
|
[Ray wait] nodes=14/14 GPUs=56.0/56.0 resources={'memory': 10624069140480.0, 'GPU': 56.0, 'node:__internal_head__': 1.0, 'object_store_memory': 601295421440.0, 'node:10.128.43.81': 1.0, 'CPU': 4032.0, 'accelerator_type:GH200': 14.0, 'node:10.128.43.96': 1.0, 'node:10.128.43.84': 1.0, 'node:10.128.43.90': 1.0, 'node:10.128.43.103': 1.0, 'node:10.128.43.85': 1.0, 'node:10.128.43.86': 1.0, 'node:10.128.43.95': 1.0, 'node:10.128.43.87': 1.0, 'node:10.128.43.93': 1.0, 'node:10.128.43.94': 1.0, 'node:10.128.43.88': 1.0, 'node:10.128.43.91': 1.0, 'node:10.128.43.108': 1.0}
|
|||
|
|
✓ Ray cluster ready
|
|||
|
|
=== Ray Cluster Ready ===
|
|||
|
|
Address: 10.128.43.81:6379
|
|||
|
|
Total GPUs: 56
|
|||
|
|
=========================
|
|||
|
|
Ray cluster ready at 10.128.43.81:6379
|
|||
|
|
Total GPUs available: 56
|
|||
|
|
[RLJobRunner] Pinggy check: url=False, token=False, needs_tunnel=False (agent=terminus-2, env=daytona)
|
|||
|
|
[RLJobRunner] No Pinggy tunnel needed, using local vLLM
|
|||
|
|
|
|||
|
|
Running SkyRL:
|
|||
|
|
Python: /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/bin/python
|
|||
|
|
Entrypoint: examples.terminal_bench.entrypoints.main_tbench
|
|||
|
|
Args: 119 Hydra arguments
|
|||
|
|
Working dir: /e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train
|
|||
|
|
Using proxychains binary: /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4
|
|||
|
|
|
|||
|
|
Executing command with srun: /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/bin/proxychains4 -f $PROXYCHAINS_CONF_FILE /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/bin/python -m examples.terminal_bench.entrypoints.main_tbench +terminal_bench_config=terminal_bench trainer.strategy=fsdp2 trainer.algorithm.advantage_estimator=rloo_n trainer.algorithm.use_kl_loss=false trainer.algorithm.kl_loss_coef=0.0 trainer.algorithm.eps_clip_low=0.2 trainer.algorithm.eps_clip_high=0.05 trainer.algorithm.loss_reduction=token_mean trainer.epochs=2 trainer.max_steps=80 trainer.update_epochs_per_batch=1 trainer.train_batch_size=64 trainer.policy_mini_batch_size=64 trainer.eval_batch_size=64 trainer.micro_forward_batch_size_per_gpu=4 trainer.micro_train_batch_size_per_gpu=1 trainer.max_prompt_length=999999 trainer.eval_interval=999999 trainer.eval_before_train=false trainer.ckpt_interval=2 trainer.resume_mode=latest trainer.hf_save_interval=5 ++trainer.hf_hub_repo_id=laion/a3-rl-DCAgent_code-contests-noblock ++trainer.hf_hub_private=false ++trainer.hf_hub_revision=main ++trainer.enable_db_registration=true trainer.project_name=OpenThoughts-Agent trainer.log_level=INFO trainer.tracker_commit_each_step=true trainer.logger=console trainer.run_name=a3-rl-DCAgent_code-contests-noblock trainer.ckpt_path=/e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints trainer.export_path=/e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/exports trainer.policy.optimizer_config.lr=8e-6 trainer.policy.optimizer_config.weight_decay=0.0 trainer.policy.optimizer_config.adam_betas=[0.9,0.999] trainer.policy.optimizer_config.max_grad_norm=0.9 trainer.policy.fsdp_config.cpu_offload=false trainer.policy.fsdp_config.reshard_after_forward=true trainer.policy.fsdp_config.fsdp_size=4 trainer.policy.model.path=/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6 trainer.ref.fsdp_config.cpu_offload=false trainer.ref.fsdp_config.reshard_after_forward=true trainer.ref.fsdp_config.fsdp_size=4 trainer.placement.colocate_all=false trainer.placement.policy_num_nodes=2 trainer.placement.ref_num_nodes=2 trainer.placement.policy_num_gpus_per_node=4 trainer.placement.ref_num_gpus_per_node=4 trainer.fully_async.max_staleness_steps=16 trainer.fully_async.num_parallel_generation_workers=450 generator.backend=vllm generator.timeout_multiplier=1.0 generator.model_dtype=bfloat16 generator.inference_engine_tensor_parallel_size=1 generator.num_inference_engines=48 generator.n_samples_per_prompt=8 generator.eval_n_samples_per_prompt=8 generator.gpu_memory_utilization=0.75 generator.max_num_seqs=24 generator.max_num_batched_tokens=65536 generator.enable_prefix_caching=true generator.enable_chunked_prefill=true generator.run_engines_locally=true generator.weight_sync_backend=nccl generator.async_engine=true generator.batched=false generator.enable_http_endpoint=true generator.enable_ray_prometheus_stats=false generator.vllm_stats_interval=1 generator.append_eos_token_after_stop_str_in_multi_turn=true generator.max_turns=999999 generator.sampling_params.max_generate_length=4096 generator.sampling_params.temperature=0.7 generator.sampling_params.top_p=0.95 generator.sampling_params.top_k=20 ++generator.engine_init_kwargs.max_model_len=32768 ++generator.engine_init_kwargs.custom_chat_template_chat_completion_path=chat_templates/qwen3_thinking_acc.jinja2 ++generator.engine_init_kwargs.served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6 data.train_data=["/e/scratch/jureap59/feuer1/tasks/code-contests-noblock"] data.val_data=[] +terminal_bench_config.trials_dir=/e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/trace_jobs +terminal_bench_config.harbor.name=terminus-2 +terminal_bench_config.harbor.max_episodes=999999 +terminal_bench_config.harbor.enable_summarize=false +t
|
|||
|
|
[proxychains] config file found: /e/home/jusers/feuer1/jupiter/.proxychains/proxychains_485102.conf
|
|||
|
|
[proxychains] preloading /e/scratch/jureap59/feuer1/proxychains-ng-aarch64/lib/libproxychains4.so
|
|||
|
|
[proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e
|
|||
|
|
2026-05-23 01:51:48.647 | INFO | skyrl_train.utils.utils:prepare_runtime_environment:590 - Exporting wandb api key to ray runtime env
|
|||
|
|
2026-05-23 01:51:48.648 | INFO | skyrl_train.utils.utils:prepare_runtime_environment:609 - Exporting RAY_ADDRESS to ray runtime env
|
|||
|
|
2026-05-23 01:51:48,648 INFO worker.py:1680 -- Using address 10.128.43.81:6379 set in the environment variable RAY_ADDRESS
|
|||
|
|
2026-05-23 01:51:48,679 INFO worker.py:1821 -- Connecting to existing Ray cluster at address: 10.128.43.81:6379...
|
|||
|
|
2026-05-23 01:51:48,689 INFO worker.py:2007 -- Connected to Ray cluster.
|
|||
|
|
/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/worker.py:2046: FutureWarning: Tip: In future versions of Ray, Ray will no longer override accelerator visible devices env var if num_gpus=0 or num_gpus=None (default). To enable this behavior and turn off this error message, set RAY_ACCEL_ENV_VAR_OVERRIDE_ON_ZERO=0
|
|||
|
|
warnings.warn(
|
|||
|
|
[33m(raylet, ip=10.128.43.87)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e
|
|||
|
|
2026-05-23 01:51:51.188 | INFO | skyrl_train.utils.ppo_utils:sync_registries:546 - Synced registries to ray actor
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:52:01.959[0m | [1mINFO [0m | [36mskyrl_train.entrypoints.main_base[0m:[36m_configure_log_level[0m:[36m199[0m - [1mSkyRL log level set to: INFO[0m
|
|||
|
|
[33m(raylet)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 17x across cluster] (Ray deduplicates logs by default. Set RAY_DEDUP_LOGS=0 to disable log deduplication, or see https://docs.ray.io/en/master/ray-observability/user-guides/configure-logging.html#log-deduplication for more options.)[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:52:02.310[0m | [1mINFO [0m | [36mexamples.terminal_bench.dataset[0m:[36m_load_data_files[0m:[36m40[0m - [1mLoading data from: /e/scratch/jureap59/feuer1/tasks/code-contests-noblock[0m
|
|||
|
|
[2026-05-23 01:52:18,917 E 1460804 1461255] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14
|
|||
|
|
[36m(RegistryActor pid=1657870, ip=10.128.43.87)[0m [2026-05-23 01:52:19,865 E 1657870 1657910] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:53:55.602[0m | [1mINFO [0m | [36mexamples.terminal_bench.dataset[0m:[36m_load_data_files[0m:[36m50[0m - [1mFound 8728 valid task directories out of 8728 total directories[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:53:55.602[0m | [1mINFO [0m | [36mexamples.terminal_bench.dataset[0m:[36m__init__[0m:[36m27[0m - [1mTerminalBenchTaskDataset initialized with 8728 task paths[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:53:55.613[0m | [1mINFO [0m | [36mskyrl_train.entrypoints.main_base[0m:[36m_setup_trainer[0m:[36m352[0m - [1mdata:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m train_data:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - /e/scratch/jureap59/feuer1/tasks/code-contests-noblock
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m val_data: []
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m trainer:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m placement:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m colocate_all: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m colocate_policy_ref: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m policy_num_nodes: 2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m policy_num_gpus_per_node: 4
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m critic_num_nodes: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m critic_num_gpus_per_node: 4
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ref_num_nodes: 2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ref_num_gpus_per_node: 4
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m sequence_parallel_backend: ulysses
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m strategy: fsdp2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m policy:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m path: /e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m lora:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rank: 0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m alpha: 16
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dropout: 0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m lora_sync_path: /tmp/skyrl_lora_sync
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m target_modules: all-linear
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m exclude_modules: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m deepspeed_config: ${deepspeed_config.train}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer: AdamW
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m lr: 8.0e-06
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m adam_betas:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - 0.9
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - 0.999
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m weight_decay: 0.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_grad_norm: 0.9
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m offload_after_step: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m num_warmup_steps: 0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m scheduler: constant_with_warmup
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer_kwargs: {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fsdp_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cpu_offload: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m reshard_after_forward: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fsdp_size: 4
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m sequence_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_torch_compile: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m record_memory: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m megatron_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m tensor_model_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m pipeline_model_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m context_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m expert_model_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m expert_tensor_parallel_size: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ddp_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m grad_reduce_in_fp32: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m overlap_grad_reduce: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m overlap_param_gather: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m average_in_collective: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model_config_kwargs: {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m torch_profiler_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ranks: []
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m save_path: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer_config_kwargs:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m overlap_cpu_optimizer_d2h_h2d: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_precision_aware_optimizer: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer_cpu_offload: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer_offload_fraction: 0.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m transformer_config_kwargs:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m recompute_granularity: full
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m recompute_modules:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - core_attn
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m recompute_method: uniform
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m recompute_num_layers: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m empty_cuda_cache: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ref:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m path: ${trainer.policy.model.path}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m sequence_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m deepspeed_config: ${deepspeed_config.eval}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fsdp_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cpu_offload: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m reshard_after_forward: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fsdp_size: 4
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m megatron_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m tensor_model_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m pipeline_model_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m context_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m expert_model_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m expert_tensor_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model_config_kwargs: {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m transformer_config_kwargs: {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m critic:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m path: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m lora:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rank: 0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m alpha: 16
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dropout: 0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m target_modules: all-linear
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m exclude_modules: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m deepspeed_config: ${deepspeed_config.train}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer: AdamW
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m lr: 5.0e-06
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m adam_betas:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - 0.9
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - 0.999
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m weight_decay: 0.01
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_grad_norm: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m offload_after_step: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m num_warmup_steps: 0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m scheduler: constant_with_warmup
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m optimizer_kwargs: {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fsdp_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cpu_offload: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m reshard_after_forward: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fsdp_size: -1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m sequence_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m algorithm:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m advantage_estimator: rloo_n
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m kl_ctrl:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m type: fixed
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m kl_target: 0.1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m horizon: 10000
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m kl_estimator_type: k3
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_kl_estimator_k3: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_abs_kl: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_kl_in_reward: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_kl_loss: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m kl_loss_coef: 0.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_entropy_loss: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m entropy_loss_coef: 0.01
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m advantage_batch_normalize: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m value_head_prefix: value_head
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m policy_loss_type: regular
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m loss_reduction: token_mean
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m grpo_norm_by_std: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rloo_n_min_group_size: 4
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rloo_n_filter_zero_reward_groups: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m lambd: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m gamma: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eps_clip_low: 0.2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eps_clip_high: 0.05
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m clip_ratio_c: 3.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m tis_imp_ratio_cap: -1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_tis: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m sapo:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m tau_pos: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m tau_neg: 1.05
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m value_clip: 0.2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dynamic_sampling:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m type: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_sample_batches: 30
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m min_replace_ratio: 0.3
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m clip_cov:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m clip_ratio: 0.0002
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m clip_cov_lb: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m clip_cov_ub: 5.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m kl_cov:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m kl_cov_frac: 0.2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ppo_kl_coef: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cispo:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cispo_eps_clip_low: 0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cispo_eps_clip_high: 5
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m z_clip:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enabled: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m alpha: 0.97
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m z_thresh: 2.5
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m warmup_steps: 25
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m mode: zscore
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m clip_option: adaptive_scaling
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m clip_factor: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m skip_update_on_spike: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stale_clip:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enabled: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m alpha: 0.3
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m entropy_threshold: 0.15
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m entropy_window: 10
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m min_lr_scale: 0.1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_seq_len: 1004095
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fully_async:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_staleness_steps: 16
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m num_parallel_generation_workers: 450
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m gradient_checkpointing: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m gradient_checkpointing_use_reentrant: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m seed: 42
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m resume_mode: latest
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m resume_path: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ckpt_path: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_ckpts_to_keep: -1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ckpt_interval: 2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m hf_save_interval: 5
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m export_path: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/exports
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m bf16: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m epochs: 2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_steps: 80
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m update_epochs_per_batch: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m train_batch_size: 64
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m policy_mini_batch_size: 64
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m critic_mini_batch_size: 256
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m micro_train_batch_size_per_gpu: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m micro_forward_batch_size_per_gpu: 4
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m update_ref_every_epoch: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_sample_packing: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eval_batch_size: 64
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eval_before_train: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eval_interval: 999999
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_prompt_length: 999999
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m flash_attn: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m disable_fast_tokenizer: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m target_modules: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m exclude_modules: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m project_name: OpenThoughts-Agent
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m run_name: a3-rl-DCAgent_code-contests-noblock
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m logger: console
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m tracker_commit_each_step: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dump_data_batch: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dump_eval_results: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m log_level: INFO
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rope_scaling: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rope_theta: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m step_wise_training: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m hf_hub_repo_id: laion/a3-rl-DCAgent_code-contests-noblock
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m hf_hub_private: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m hf_hub_revision: main
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_db_registration: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m generator:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model_name: ${trainer.policy.model.path}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model_dtype: bfloat16
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m timeout_multiplier: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m run_engines_locally: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m num_inference_engines: 48
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m backend: vllm
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m weight_sync_backend: nccl
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fuse_weights: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m weight_transfer_threshold_cuda_ipc_GB: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m inference_engine_tensor_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m inference_engine_pipeline_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m inference_engine_expert_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m inference_engine_data_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m n_samples_per_prompt: 8
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m async_engine: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m batched: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_input_length: ${trainer.max_prompt_length}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m vllm_v1_disable_multiproc: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_prefix_caching: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_chunked_prefill: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_num_batched_tokens: 65536
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enforce_eager: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m fully_sharded_loras: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_ray_prometheus_stats: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m vllm_stats_interval: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m gpu_memory_utilization: 0.75
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_num_seqs: 24
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m remote_inference_engine_urls:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - 127.0.0.1:8001
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_http_endpoint: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m http_endpoint_host: 127.0.0.1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m http_endpoint_port: 8000
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_turns: 999999
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m chat_template:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m source: name
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m name_or_path: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m chat_template_kwargs: {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m engine_init_kwargs:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_model_len: 32768
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m custom_chat_template_chat_completion_path: chat_templates/qwen3_thinking_acc.jinja2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m served_model_name: 0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m override_existing_update_group: disable
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m sampling_params:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_generate_length: 4096
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m repetition_penalty: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m temperature: 0.7
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m top_p: 0.95
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m min_p: 0.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m top_k: 20
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m logprobs: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stop: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m use_conversation_multi_turn: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m append_eos_token_after_stop_str_in_multi_turn: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eval_sampling_params:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_generate_length: ${generator.sampling_params.max_generate_length}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m repetition_penalty: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m temperature: 0.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m top_p: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m min_p: 0.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m top_k: -1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m logprobs: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stop: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eval_n_samples_per_prompt: 8
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m zero_reward_on_non_stop: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m apply_overlong_filtering: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rope_scaling: ${trainer.rope_scaling}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rope_theta: ${trainer.rope_theta}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m teacher:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model_path: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m top_k_logprobs: 256
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m num_inference_engines: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m inference_engine_tensor_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m inference_engine_pipeline_parallel_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m gpu_memory_utilization: 0.9
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enforce_eager: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m backend: vllm
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m engine_init_kwargs: {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m environment:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m env_class: gsm8k
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m skyrl_gym:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_env_workers: 32
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m text2sql:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m db_path: /home/ray/default/sql_data
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m llm_as_a_judge:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model: gpt-4o-mini
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m base_url: null
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m search:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m log_requests: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m search_url: http://127.0.0.1:8000/retrieve
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m topk: 3
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m timeout: 30
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m deepspeed_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m train:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m zero_optimization:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stage: 3
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m offload_param:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m device: none
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m offload_optimizer:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m device: none
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m pin_memory: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m sub_group_size: auto
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m reduce_bucket_size: auto
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stage3_param_persistence_threshold: auto
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stage3_prefetch_bucket_size: auto
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stage3_max_live_parameters: auto
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stage3_max_reuse_distance: auto
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m round_robin_gradients: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m zero_hpz_partition_size: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m zero_quantized_weights: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m zero_quantized_gradients: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m torch_autocast:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enabled: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dtype: bfloat16
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m disable_trace_cache: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m data_types:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m grad_accum_dtype: fp32
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m gradient_clipping: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m wall_clock_breakdown: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m prescale_gradient: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m eval:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m zero_optimization:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stage: 3
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m stage3_param_persistence_threshold: auto
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m offload_param:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m device: cpu
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m pin_memory: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m torch_autocast:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enabled: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dtype: bfloat16
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m gradient_clipping: 1.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m prescale_gradient: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m wall_clock_breakdown: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m terminal_bench_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m trials_dir: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/trace_jobs
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m harbor:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m name: terminus-2
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_episodes: 999999
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_summarize: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m store_all_messages: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m trajectory_config:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m raw_content: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_episode_logging: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m record_terminal_session: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_pane_logging: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m strict_json_parser: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m interleaved_thinking: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m extra_body:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m chat_template_kwargs:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_thinking: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m override_timeout_sec: 1800
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m override_cpus: 1
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m override_memory_mb: 2048
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m override_storage_mb: 2048
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m auto_snapshot: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m verifier_override_timeout_sec: 120
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_retries: 3
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m min_wait_sec: 60.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_wait_sec: 600.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m wait_multiplier: 2.0
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m exclude_exceptions:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - VerifierTimeoutError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - VerifierRuntimeError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - RewardFileNotFoundError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - RewardFileEmptyError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - VerifierOutputParseError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m n_concurrent_trials: 900
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m log_level: INFO
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_reward_shaping: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enable_error_classification: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m mask_exceptions:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - DaytonaError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - EnvironmentStartTimeoutError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - NetworkError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - ConnectionError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - RewardFileNotFoundError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - RewardFileEmptyError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - AgentEnvironmentTimeoutError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m default_error_treatment: zero
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m passthrough_exceptions:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - AgentTimeoutError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m - ContextLengthExceededError
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m model_info:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_input_tokens: 32768
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m max_output_tokens: 4096
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m archiving:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enabled: false
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m trace_upload:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m enabled: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m repo_org: DCAgent
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m episodes: last
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m dataset_type: SFT
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cleanup: true
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [2026-05-23 01:52:22,342 E 1461281 1461324] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14[32m [repeated 2x across cluster][0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m No module named 'vllm._version'
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m from .version import __version__, __version_tuple__ # isort:skip
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m W0523 01:54:12.721000 1461281 envs/rl/lib/python3.12/site-packages/torch/utils/cpp_extension.py:117] No CUDA runtime is found, using CUDA_HOME='/e/software/default/stages/2026/software/CUDA/13'
|
|||
|
|
[33m(raylet)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e
|
|||
|
|
[33m(raylet, ip=10.128.43.84)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 41x across cluster][0m
|
|||
|
|
[36m(pid=873920, ip=10.128.43.93)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:
|
|||
|
|
[36m(pid=873920, ip=10.128.43.93)[0m No module named 'vllm._version'
|
|||
|
|
[36m(pid=873920, ip=10.128.43.93)[0m from .version import __version__, __version_tuple__ # isort:skip
|
|||
|
|
[33m(raylet, ip=10.128.43.108)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 98x across cluster][0m
|
|||
|
|
[36m(pid=874053, ip=10.128.43.93)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(pid=874053, ip=10.128.43.93)[0m No module named 'vllm._version'[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(pid=874053, ip=10.128.43.93)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 15x across cluster][0m
|
|||
|
|
[33m(raylet, ip=10.128.43.95)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 72x across cluster][0m
|
|||
|
|
[36m(pid=900781, ip=10.128.43.91)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(pid=900781, ip=10.128.43.91)[0m No module named 'vllm._version'[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(pid=900781, ip=10.128.43.91)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:35.893 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:123 - setup_envvars_for_vllm: distributed_executor_backend=uni, SKYRL_ENABLE_NUMA_AFFINITY=1, CUDA_VISIBLE_DEVICES=0, VLLM_ENABLE_V1_MULTIPROCESSING=<unset>
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:35.895 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:133 - setup_envvars_for_vllm: numa_enabled=True
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:35.895 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:136 - setup_envvars_for_vllm: set VLLM_ENABLE_V1_MULTIPROCESSING=0 for NUMA affinity
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:37.262 | INFO | skyrl_train.utils.numa:set_numa_affinity_for_gpu:340 - NUMA affinity: bound process to CPUs 0-71 (NUMA node 0) for GPU 0
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:37.280 | DEBUG | skyrl_train.utils.numa:_set_membind_via_libnuma:375 - NUMA affinity: set memory preferred to NUMA node 0
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:37.280 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:481 - BaseVLLMInferenceEngine: vllm_v1_disable_multiproc=True, vllm.__version__=dev, VLLM_ENABLE_V1_MULTIPROCESSING=0
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:37.280 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:489 - BaseVLLMInferenceEngine: set VLLM_ENABLE_V1_MULTIPROCESSING=0
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:37.280 | WARNING | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1047 - OpenAI API sampling params overridden: temperature=0.7, top_p=0.95, top_k=20
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m 2026-05-23 01:54:37.311 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1078 - Engine startup stagger: sleeping 2.82s to avoid port collisions
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.
|
|||
|
|
[33m(raylet, ip=10.128.43.90)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 106x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874054, ip=10.128.43.93)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 18x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874054, ip=10.128.43.93)[0m No module named 'vllm._version'[32m [repeated 18x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874054, ip=10.128.43.93)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 18x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m 2026-05-23 01:54:39.678 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:123 - setup_envvars_for_vllm: distributed_executor_backend=uni, SKYRL_ENABLE_NUMA_AFFINITY=1, CUDA_VISIBLE_DEVICES=3, VLLM_ENABLE_V1_MULTIPROCESSING=<unset>[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m 2026-05-23 01:54:39.678 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:133 - setup_envvars_for_vllm: numa_enabled=True[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m 2026-05-23 01:54:39.678 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:136 - setup_envvars_for_vllm: set VLLM_ENABLE_V1_MULTIPROCESSING=0 for NUMA affinity[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [2026-05-23 01:54:41] INFO inference_engine_client_http_endpoint.py:350: Starting server on 127.0.0.1:8000
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [2026-05-23 01:54:41] INFO inference_engine_client_http_endpoint.py:242: Starting inference HTTP endpoint...
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [2026-05-23 01:54:42] INFO inference_engine_client_http_endpoint.py:229: Server ready after 2 attempts (2 seconds)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.194[0m | [1mINFO [0m | [36mskyrl_train.inference_engines.inference_engine_client[0m:[36m_spin_up_http_endpoint[0m:[36m960[0m - [1mInferenceEngineClient HTTP endpoint started on 127.0.0.1:8000[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.194[0m | [1mINFO [0m | [36mskyrl_train.inference_engines.inference_engine_client[0m:[36m__init__[0m:[36m61[0m - [1mInferenceEngineClient initialized with 48 engines.[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.196[0m | [1mINFO [0m | [36mexamples.terminal_bench.terminal_bench_generator[0m:[36m_configure_harbor_logging[0m:[36m213[0m - [1mHarbor logging level set to INFO[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.197[0m | [1mINFO [0m | [36mexamples.terminal_bench.terminal_bench_generator[0m:[36m__init__[0m:[36m113[0m - [1mTerminalBenchGenerator initialized with HarborConfigBuilder. Exposed fields: ['name', 'max_episodes', 'enable_summarize', 'store_all_messages', 'trajectory_config', 'enable_episode_logging', 'record_terminal_session', 'enable_pane_logging', 'strict_json_parser', 'interleaved_thinking', 'extra_body', 'override_timeout_sec', 'override_cpus', 'override_memory_mb', 'override_storage_mb', 'auto_snapshot', 'verifier_override_timeout_sec', 'max_retries', 'min_wait_sec', 'max_wait_sec', 'wait_multiplier', 'exclude_exceptions', 'n_concurrent_trials', 'log_level', 'enable_reward_shaping', 'enable_error_classification', 'mask_exceptions', 'default_error_treatment', 'passthrough_exceptions']. Retry config: max_retries=3, backoff=60.0-600.0s. Concurrent trials: 900. Reward shaping: enabled=False, shaper=pass_ratio. Error classification: enabled=True[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.210[0m | [1mINFO [0m | [36mexamples.terminal_bench.terminal_bench_generator[0m:[36m__init__[0m:[36m129[0m - [1mTerminalBenchGenerator initialized with custom chat template read from: chat_templates/qwen3_thinking_acc.jinja2[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.211[0m | [1mINFO [0m | [36mskyrl_train.utils.trainer_utils[0m:[36mbuild_dataloader[0m:[36m656[0m - [1mTotal steps: 272[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.211[0m | [1mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_build_train_dataloader_and_compute_training_steps[0m:[36m351[0m - [1mLength of train_dataloader: 8728[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.211[0m | [1mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_build_train_dataloader_and_compute_training_steps[0m:[36m352[0m - [1mNumber of steps per epoch: 136[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:42.211[0m | [1mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_build_train_dataloader_and_compute_training_steps[0m:[36m353[0m - [1mTotal training steps: 80[0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m 2026-05-23 01:54:41.065 | INFO | skyrl_train.utils.numa:set_numa_affinity_for_gpu:340 - NUMA affinity: bound process to CPUs 216-287 (NUMA node 3) for GPU 3[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658031, ip=10.128.43.87)[0m 2026-05-23 01:54:42.409 | DEBUG | skyrl_train.utils.numa:_set_membind_via_libnuma:375 - NUMA affinity: set memory preferred to NUMA node 0[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658031, ip=10.128.43.87)[0m 2026-05-23 01:54:42.409 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:481 - BaseVLLMInferenceEngine: vllm_v1_disable_multiproc=True, vllm.__version__=dev, VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658031, ip=10.128.43.87)[0m 2026-05-23 01:54:42.409 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:489 - BaseVLLMInferenceEngine: set VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658031, ip=10.128.43.87)[0m 2026-05-23 01:54:42.409 | WARNING | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1047 - OpenAI API sampling params overridden: temperature=0.7, top_p=0.95, top_k=20[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m 2026-05-23 01:54:41.128 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1078 - Engine startup stagger: sleeping 2.24s to avoid port collisions[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 69x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937426, ip=10.128.43.84)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 14x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937426, ip=10.128.43.84)[0m No module named 'vllm._version'[32m [repeated 14x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937426, ip=10.128.43.84)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 14x across cluster][0m
|
|||
|
|
[36m(pid=851205, ip=10.128.43.85)[0m Using blocking ray.get inside async actor. This blocks the event loop. Please use `await` on object ref with asyncio.gather if you want to yield execution to the event loop instead.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m 2026-05-23 01:54:45.957 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:123 - setup_envvars_for_vllm: distributed_executor_backend=uni, SKYRL_ENABLE_NUMA_AFFINITY=1, CUDA_VISIBLE_DEVICES=2, VLLM_ENABLE_V1_MULTIPROCESSING=<unset>[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m 2026-05-23 01:54:45.958 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:133 - setup_envvars_for_vllm: numa_enabled=True[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m 2026-05-23 01:54:45.958 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:136 - setup_envvars_for_vllm: set VLLM_ENABLE_V1_MULTIPROCESSING=0 for NUMA affinity[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(bundle_reservation_check_func pid=1461399)[0m [2026-05-23 01:54:47,098 E 1461399 1461441] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m 2026-05-23 01:54:47.356 | INFO | skyrl_train.utils.numa:set_numa_affinity_for_gpu:340 - NUMA affinity: bound process to CPUs 216-287 (NUMA node 3) for GPU 3[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m 2026-05-23 01:54:47.380 | DEBUG | skyrl_train.utils.numa:_set_membind_via_libnuma:375 - NUMA affinity: set memory preferred to NUMA node 3[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m 2026-05-23 01:54:47.380 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:481 - BaseVLLMInferenceEngine: vllm_v1_disable_multiproc=True, vllm.__version__=dev, VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m 2026-05-23 01:54:47.380 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:489 - BaseVLLMInferenceEngine: set VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m 2026-05-23 01:54:47.380 | WARNING | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1047 - OpenAI API sampling params overridden: temperature=0.7, top_p=0.95, top_k=20[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m 2026-05-23 01:54:47.403 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1078 - Engine startup stagger: sleeping 2.57s to avoid port collisions[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.[32m [repeated 30x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900916, ip=10.128.43.91)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 32x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m No module named 'vllm._version'[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:49.703 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:123 - setup_envvars_for_vllm: distributed_executor_backend=uni, SKYRL_ENABLE_NUMA_AFFINITY=1, CUDA_VISIBLE_DEVICES=3, VLLM_ENABLE_V1_MULTIPROCESSING=<unset>[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:49.704 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:133 - setup_envvars_for_vllm: numa_enabled=True[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:49.704 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:136 - setup_envvars_for_vllm: set VLLM_ENABLE_V1_MULTIPROCESSING=0 for NUMA affinity[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874281) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/__init__.py:1617: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874281) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937299, ip=10.128.43.84)[0m [2026-05-23 01:54:51,552 E 937299 937401] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:50.968 | INFO | skyrl_train.utils.numa:set_numa_affinity_for_gpu:340 - NUMA affinity: bound process to CPUs 216-287 (NUMA node 3) for GPU 3[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:50.989 | DEBUG | skyrl_train.utils.numa:_set_membind_via_libnuma:375 - NUMA affinity: set memory preferred to NUMA node 3[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:50.989 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:481 - BaseVLLMInferenceEngine: vllm_v1_disable_multiproc=True, vllm.__version__=dev, VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:50.989 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:489 - BaseVLLMInferenceEngine: set VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:50.990 | WARNING | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1047 - OpenAI API sampling params overridden: temperature=0.7, top_p=0.95, top_k=20[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:54:51.045 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1078 - Engine startup stagger: sleeping 2.40s to avoid port collisions[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874052, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874286) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m [W523 01:54:54.227983559 socket.cpp:767] [c10d] The client socket cannot be initialized to connect to [jpbo-078-45.jupiter.internal]:49029 (errno: 97 - Address family not supported by protocol).
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m [W523 01:54:54.231619642 Utils.hpp:166] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator())
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m [rank0]:[W523 01:54:54.236680175 Utils.hpp:137] Warning: Environment variable TORCH_NCCL_TRACE_BUFFER_SIZE is deprecated; use TORCH_FR_BUFFER_SIZE instead (function operator())
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m 2026-05-23 01:54:54 INFO [ipv4-debug] hostname=jpbo-078-37.jupiter.internal
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m 2026-05-23 01:54:54 INFO [ipv4-debug] _global_node.node_ip_address=10.128.43.85
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m 2026-05-23 01:54:54 INFO [ipv4-debug] get_node_ip_address()=10.128.43.85
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.[32m [repeated 16x across cluster][0m
|
|||
|
|
[33m(raylet)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 20x across cluster][0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:54:54.377[0m | [1m[32mINFO [0m | [36mskyrl_train.workers.worker[0m:[36m_initiate_actors[0m:[36m496[0m - [1m[32mInitializing process group for RayActorGroup[0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874054, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874292) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937427, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937657) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m No module named 'vllm._version'[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937670) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937299, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937662) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:56.483 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:123 - setup_envvars_for_vllm: distributed_executor_backend=uni, SKYRL_ENABLE_NUMA_AFFINITY=1, CUDA_VISIBLE_DEVICES=3, VLLM_ENABLE_V1_MULTIPROCESSING=<unset>[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:56.484 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:133 - setup_envvars_for_vllm: numa_enabled=True[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:56.484 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:setup_envvars_for_vllm:136 - setup_envvars_for_vllm: set VLLM_ENABLE_V1_MULTIPROCESSING=0 for NUMA affinity[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658393) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307)
Loading safetensors checkpoint shards: 0% Completed | 0/4 [00:00<?, ?it/s]
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=839106, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839345) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658031, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658398) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(pid=851282, ip=10.128.43.85)[0m Using blocking ray.get inside async actor. This blocks the event loop. Please use `await` on object ref with asyncio.gather if you want to yield execution to the event loop instead.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876508, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876740) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937426, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937681) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/__init__.py:1617: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.)[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m [2026-05-23 01:54:56,430 E 876379 876482] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839337) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=839108, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839350) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658164, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658411) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:57.854 | INFO | skyrl_train.utils.numa:set_numa_affinity_for_gpu:340 - NUMA affinity: bound process to CPUs 216-287 (NUMA node 3) for GPU 3[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:57.879 | DEBUG | skyrl_train.utils.numa:_set_membind_via_libnuma:375 - NUMA affinity: set memory preferred to NUMA node 3[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:57.879 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:481 - BaseVLLMInferenceEngine: vllm_v1_disable_multiproc=True, vllm.__version__=dev, VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:57.879 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:__init__:489 - BaseVLLMInferenceEngine: set VLLM_ENABLE_V1_MULTIPROCESSING=0[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:57.879 | WARNING | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1047 - OpenAI API sampling params overridden: temperature=0.7, top_p=0.95, top_k=20[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m 2026-05-23 01:54:57.910 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1078 - Engine startup stagger: sleeping 2.00s to avoid port collisions[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=839107, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839358) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876509, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876761) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) Process EngineCore_DP0:
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) Traceback (most recent call last):
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/multiprocessing/process.py", line 314, in _bootstrap
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) self.run()
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/multiprocessing/process.py", line 108, in run
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) self._target(*self._args, **self._kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 1010, in run_engine_core
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) raise e
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 996, in run_engine_core
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) engine_core = EngineCoreProc(*args, engine_index=dp_rank, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 740, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) super().__init__(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 106, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) self.model_executor = executor_class(vllm_config)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/executor/abstract.py", line 103, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) self._init_executor()
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/executor/uniproc_executor.py", line 47, in _init_executor
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) self.driver_worker.init_device()
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/worker/worker_base.py", line 332, in init_device
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) self.worker.init_device() # type: ignore
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/worker/gpu_worker.py", line 232, in init_device
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) init_worker_distributed_environment(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/worker/gpu_worker.py", line 1040, in init_worker_distributed_environment
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) init_distributed_environment(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/distributed/parallel_state.py", line 1256, in init_distributed_environment
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) torch.distributed.init_process_group(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) return func(*args, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/c10d_logger.py", line 95, in wrapper
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) func_return = func(*args, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py", line 1762, in init_process_group
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) store, rank, world_size = next(rendezvous_iterator)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/rendezvous.py", line 230, in _tcp_rendezvous_handler
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) store = _create_c10d_store(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/rendezvous.py", line 198, in _create_c10d_store
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) return TCPStore(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) torch.distributed.DistNetworkError: The server socket has failed to listen on any local network address. port: 36099, useIpv6: false, code: -98, name: EADDRINUSE, message: address already in use
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) Exception raised from makeWithPort at /pytorch/torch/csrc/distributed/c10d/TCPStoreLibUvBackend.cpp:307 (most recent call first):
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) + 0xb0 (0x40004b8fc700 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libc10.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #1: <unknown function> + 0x5f29220 (0x40002ba99220 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #2: <unknown function> + 0x5f4326c (0x40002bab326c in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #3: <unknown function> + 0x5f49074 (0x40002bab9074 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #4: <unknown function> + 0x5f49138 (0x40002bab9138 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #5: <unknown function> + 0x5f2ccc4 (0x40002ba9ccc4 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #6: c10d::TCPStore::TCPStore(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, c10d::TCPStoreOptions const&) + 0x104 (0x40002baa1564 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #7: <unknown function> + 0x109a094 (0x40002584a094 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #8: <unknown function> + 0x113236c (0x4000258e236c in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #9: <unknown function> + 0x5d6d60 (0x400024d86d60 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #10: <unknown function> + 0x1b7a38 (0xaaaab65b7a38 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #11: _PyObject_MakeTpCall + 0x98 (0xaaaab6565db8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #12: <unknown function> + 0x169f50 (0xaaaab6569f50 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #13: <unknown function> + 0x1682e4 (0xaaaab65682e4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #14: <unknown function> + 0x1e0ce8 (0xaaaab65e0ce8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #15: <unknown function> + 0x1d7ddc (0xaaaab65d7ddc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #16: <unknown function> + 0x646b0c (0x400024df6b0c in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #17: _PyObject_MakeTpCall + 0x98 (0xaaaab6565db8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #18: _PyEval_EvalFrameDefault + 0x280c (0xaaaab666ae54 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #19: <unknown function> + 0x1808c0 (0xaaaab65808c0 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #20: <unknown function> + 0x182bf8 (0xaaaab6582bf8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #21: <unknown function> + 0x25fd30 (0xaaaab665fd30 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #22: <unknown function> + 0x1b7d20 (0xaaaab65b7d20 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #23: PyObject_Vectorcall + 0x54 (0xaaaab65660e4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #24: _PyEval_EvalFrameDefault + 0x280c (0xaaaab666ae54 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #25: _PyObject_FastCallDictTstate + 0x80 (0xaaaab6567f20 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #26: _PyObject_Call_Prepend + 0x140 (0xaaaab65681ec in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #27: <unknown function> + 0x1e0d80 (0xaaaab65e0d80 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #28: <unknown function> + 0x1d7ddc (0xaaaab65d7ddc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #29: _PyObject_MakeTpCall + 0x98 (0xaaaab6565db8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #30: _PyEval_EvalFrameDefault + 0x280c (0xaaaab666ae54 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #31: _PyObject_FastCallDictTstate + 0x10c (0xaaaab6567fac in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #32: _PyObject_Call_Prepend + 0x140 (0xaaaab65681ec in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #33: <unknown function> + 0x1e0d80 (0xaaaab65e0d80 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #34: <unknown function> + 0x1d7ddc (0xaaaab65d7ddc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #35: _PyObject_Call + 0x68 (0xaaaab6568488 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #36: _PyEval_EvalFrameDefault + 0x52ac (0xaaaab666d8f4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #37: PyEval_EvalCode + 0xb4 (0xaaaab6672eb4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #38: <unknown function> + 0x2ccdcc (0xaaaab66ccdcc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #39: <unknown function> + 0x2ccef4 (0xaaaab66ccef4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #40: PyRun_StringFlags + 0x90 (0xaaaab66d1050 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #41: PyRun_SimpleStringFlags + 0x58 (0xaaaab66d10f8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #42: Py_RunMain + 0x2c8 (0xaaaab66f9190 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #43: Py_BytesMain + 0x64 (0xaaaab66f9fb4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #44: <unknown function> + 0x27540 (0x400013df7540 in /lib64/libc.so.6)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #45: __libc_start_main + 0x98 (0x400013df7618 in /lib64/libc.so.6)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) frame #46: <unknown function> + 0x10e0c0 (0xaaaab650e0c0 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876509, ip=10.128.43.96)[0m [W523 01:54:59.553303224 socket.cpp:767] [c10d] The client socket cannot be initialized to connect to [jpbo-078-48.jupiter.internal]:54877 (errno: 97 - Address family not supported by protocol).[32m [repeated 18x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876509, ip=10.128.43.96)[0m [W523 01:54:59.553793743 Utils.hpp:166] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator())[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876509, ip=10.128.43.96)[0m [rank0]:[W523 01:54:59.556104773 Utils.hpp:137] Warning: Environment variable TORCH_NCCL_TRACE_BUFFER_SIZE is deprecated; use TORCH_FR_BUFFER_SIZE instead (function operator())[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.[32m [repeated 20x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 58x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m No module named 'vllm._version'[32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900915, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901149) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307)
Loading safetensors checkpoint shards: 25% Completed | 1/4 [00:04<00:12, 4.32s/it]
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m Exception raised in creation task: The actor died because of an error raised in its creation task, [36mray::AsyncVLLMInferenceEngine.__init__()[39m (pid=876507, ip=10.128.43.96, actor_id=58494295fdaf9ec8226b54f502000000, repr=<skyrl_train.inference_engines.vllm.vllm_engine.AsyncVLLMInferenceEngine object at 0x400be192c230>)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 456, in result
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m return self.__get_result()
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m raise self._exception
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1028, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m super().__init__(*args, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 500, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m self.llm = self._create_engine(*args, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1081, in _create_engine
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m engine = vllm.AsyncLLMEngine.from_engine_args(engine_args, stat_loggers=stat_loggers)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 251, in from_engine_args
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m return cls(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 148, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m self.engine_core = EngineCoreClient.make_async_mp_client(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 124, in make_async_mp_client
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m return AsyncMPClient(*client_args)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 835, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m super().__init__(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 490, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m with launch_core_engines(vllm_config, executor_class, log_stats) as (
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/contextlib.py", line 144, in __exit__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m next(self.gen)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 936, in launch_core_engines
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m wait_for_engine_startup(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 995, in wait_for_engine_startup
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m raise RuntimeError(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m RuntimeError: Engine core initialization failed. See root cause above. Failed core proc(s): {}
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003267) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753)
Loading safetensors checkpoint shards: 0% Completed | 0/4 [00:00<?, ?it/s][32m [repeated 18x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901162) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(pid=1461509)[0m Using blocking ray.get inside async actor. This blocks the event loop. Please use `await` on object ref with asyncio.gather if you want to yield execution to the event loop instead.[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003033, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003273) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003033, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003273) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/__init__.py:1617: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.)[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m [2026-05-23 01:55:02,094 E 2095386 2095488] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900916, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901154) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003034, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003287) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003034, ip=10.128.43.108)[0m [W523 01:55:03.708851322 socket.cpp:767] [c10d] The client socket cannot be initialized to connect to [jpbo-079-12.jupiter.internal]:56203 (errno: 97 - Address family not supported by protocol).[32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003034, ip=10.128.43.108)[0m [W523 01:55:03.709312144 Utils.hpp:166] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator())[32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003034, ip=10.128.43.108)[0m [rank0]:[W523 01:55:03.711459426 Utils.hpp:137] Warning: Environment variable TORCH_NCCL_TRACE_BUFFER_SIZE is deprecated; use TORCH_FR_BUFFER_SIZE instead (function operator())[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1002904, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003263) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m The argument `trust_remote_code` is to be used with Auto classes. It has no effect here and is ignored.[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m [proxychains] DLL init: proxychains-ng 4.17-git-9-g78ead0e[32m [repeated 14x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m No module named 'vllm._version'[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003267)
Loading safetensors checkpoint shards: 25% Completed | 1/4 [00:01<00:05, 1.83s/it][32m [repeated 26x across cluster][0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:06.182[0m | [1m[32mINFO [0m | [36mskyrl_train.workers.worker[0m:[36m_initiate_actors[0m:[36m498[0m - [1m[32mInitialized process group for RayActorGroup[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:06.191[0m | [1m[32mINFO [0m | [36mskyrl_train.workers.worker[0m:[36m_initiate_actors[0m:[36m500[0m - [1m[32mMesh Ranks: [MeshRank(dp=0, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1), MeshRank(dp=1, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1), MeshRank(dp=2, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1), MeshRank(dp=3, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1), MeshRank(dp=4, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1), MeshRank(dp=5, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1), MeshRank(dp=6, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1), MeshRank(dp=7, sp=0, tp=0, pp=0, world_size=8, dp_size=8, pp_size=1)][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461510)[0m `torch_dtype` is deprecated! Use `dtype` instead!
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m
Loading checkpoint shards: 0%| | 0/4 [00:00<?, ?it/s]
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m
Loading checkpoint shards: 50%|█████ | 2/4 [00:00<00:00, 9.42it/s]
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1002904, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003263)
Loading safetensors checkpoint shards: 0% Completed | 0/4 [00:00<?, ?it/s][32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095513, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095757) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552032, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552262) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552032, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552262) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/__init__.py:1617: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.)[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m [2026-05-23 01:55:06,382 E 940292 940457] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095768) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940291, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940522) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461510)[0m
Loading checkpoint shards: 100%|██████████| 4/4 [00:01<00:00, 2.50it/s]
Loading checkpoint shards: 100%|██████████| 4/4 [00:01<00:00, 2.83it/s]
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940164, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940526) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095747) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095751) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m [W523 01:55:09.885732158 socket.cpp:767] [c10d] The client socket cannot be initialized to connect to [jpbo-078-38.jupiter.internal]:46833 (errno: 97 - Address family not supported by protocol).[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m [W523 01:55:09.886241428 Utils.hpp:166] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator())[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m [rank0]:[W523 01:55:09.888544871 Utils.hpp:137] Warning: Environment variable TORCH_NCCL_TRACE_BUFFER_SIZE is deprecated; use TORCH_FR_BUFFER_SIZE instead (function operator())[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876508, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876740)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876509, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876761)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2551904, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552263) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940293, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940540) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552034, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552274) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874052, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874286)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874054, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874292)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874281)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1002904, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003263)
Loading safetensors checkpoint shards: 75% Completed | 3/4 [00:05<00:02, 2.18s/it][32m [repeated 45x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869188) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851281, ip=10.128.43.85)[0m `torch_dtype` is deprecated! Use `dtype` instead![32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461512)[0m
Loading checkpoint shards: 0%| | 0/4 [00:00<?, ?it/s][32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869183) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851281, ip=10.128.43.85)[0m
Loading checkpoint shards: 75%|███████▌ | 3/4 [00:00<00:00, 3.08it/s][32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267)
Loading safetensors checkpoint shards: 0% Completed | 0/4 [00:00<?, ?it/s][32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1002904, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003263)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869183) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/__init__.py:1617: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.)[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m [2026-05-23 01:55:12,049 E 975888 976055] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003033, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003273)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003034, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003287)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003032, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003267)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868823, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869195) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839337)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=839106, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839345)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=839107, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839358)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=839108, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839350)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937299, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937662)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937426, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937681)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937427, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937657)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937554, ip=10.128.43.84)[0m (EngineCore_DP0 pid=937670)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851281, ip=10.128.43.85)[0m
Loading checkpoint shards: 100%|██████████| 4/4 [00:01<00:00, 2.30it/s]
Loading checkpoint shards: 100%|██████████| 4/4 [00:01<00:00, 2.73it/s][32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975889, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976138) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869207) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m 2026-05-23 01:55:13.517 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1115 - Initializing OpenAIServingChat with custom_chat_template read from: chat_templates/qwen3_thinking_acc.jinja2
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976133) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658031, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658398)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658393)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658164, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658411)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117) _C._set_float32_matmul_precision(precision)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m [W523 01:55:14.722835558 socket.cpp:767] [c10d] The client socket cannot be initialized to connect to [jpbo-079-07.jupiter.internal]:43683 (errno: 97 - Address family not supported by protocol).[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m [W523 01:55:14.723282685 Utils.hpp:166] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator())[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m [rank0]:[W523 01:55:14.727560583 Utils.hpp:137] Warning: Environment variable TORCH_NCCL_TRACE_BUFFER_SIZE is deprecated; use TORCH_FR_BUFFER_SIZE instead (function operator())[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531)
Loading safetensors checkpoint shards: 25% Completed | 1/4 [00:04<00:12, 4.11s/it][32m [repeated 63x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122)
Loading safetensors checkpoint shards: 0% Completed | 0/4 [00:00<?, ?it/s][32m [repeated 14x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/__init__.py:1617: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.)[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [2026-05-23 01:55:13,239 E 851205 851247] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900916, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901154)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901162)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900915, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901149)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658393) The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m 2026-05-23 01:55:18.058 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1115 - Initializing OpenAIServingChat with custom_chat_template read from: chat_templates/qwen3_thinking_acc.jinja2[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m [W523 01:55:14.098232925 socket.cpp:767] [c10d] The client socket cannot be initialized to connect to [jpbo-079-07.jupiter.internal]:58483 (errno: 97 - Address family not supported by protocol).
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m [W523 01:55:14.098713651 Utils.hpp:166] Warning: Environment variable NCCL_BLOCKING_WAIT is deprecated; use TORCH_NCCL_BLOCKING_WAIT instead (function operator())
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461509)[0m [rank7]:[W523 01:55:18.669215806 Utils.hpp:137] Warning: Environment variable TORCH_NCCL_TRACE_BUFFER_SIZE is deprecated; use TORCH_FR_BUFFER_SIZE instead (function operator())[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.565[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mbuild_models[0m:[36m802[0m - [1m[32minit policy/ref/critic models done[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.567[0m | [1m[32mINFO [0m | [36mharbor.orchestrators.queue[0m:[36mstart[0m:[36m262[0m - [1m[32m[terminal_bench_generator:236] Started 900 workers (status every 120.0s, 0.75s launch grace period)[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.567[0m | [1m[32mINFO [0m | [36mexamples.terminal_bench.terminal_bench_generator[0m:[36m_create_orchestrator[0m:[36m257[0m - [1m[32mQueueOrchestrator created and started with n_concurrent_trials=900, rollback_hook registered for ContextLengthExceededError/AgentTimeoutError[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.568[0m | [1m[32mINFO [0m | [36mexamples.terminal_bench.terminal_bench_generator[0m:[36mstartup[0m:[36m226[0m - [1m[32mTerminalBenchGenerator startup complete. Shared orchestrator ready with n_concurrent_trials=900[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.568[0m | [1m[32mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36mtrain[0m:[36m428[0m - [1m[32mGenerator startup complete[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.568[0m | [1m[32mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_train_loop[0m:[36m453[0m - [1m[32mStarted: 'load_checkpoints'[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.694[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mload_checkpoints[0m:[36m1595[0m - [1m[32mLoading checkpoint from: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints/global_step_18[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.694[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mload_checkpoints[0m:[36m1601[0m - [1m[32mResuming from global_step: 18[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.727[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mload_checkpoints[0m:[36m1617[0m - [1m[32mSuccessfully loaded trainer state[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.761[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mload_checkpoints[0m:[36m1627[0m - [1m[32mSuccessfully loaded dataloader state[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:20.762[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mload_checkpoints[0m:[36m1636[0m - [1m[32mLoading policy checkpoint from /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints/global_step_18/policy[0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531)
Loading safetensors checkpoint shards: 50% Completed | 2/4 [00:08<00:08, 4.13s/it][32m [repeated 32x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m 2026-05-23 01:55:21.838 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1115 - Initializing OpenAIServingChat with custom_chat_template read from: chat_templates/qwen3_thinking_acc.jinja2[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095751)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095513, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095757)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095768)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095747)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2551904, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552263)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552034, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552274)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552032, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552262)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267)
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851282, ip=10.128.43.85)[0m [2026-05-23 01:55:25,133 E 851282 851395] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940164, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940526)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940291, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940522)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940293, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940540)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868823, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869195)
Loading safetensors checkpoint shards: 75% Completed | 3/4 [00:12<00:04, 4.18s/it][32m [repeated 45x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2551904, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552263) The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m 2026-05-23 01:55:28.153 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1115 - Initializing OpenAIServingChat with custom_chat_template read from: chat_templates/qwen3_thinking_acc.jinja2[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868823, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869195)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869207)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869183)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868951, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869188)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975889, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976138)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976133)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122)
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461509)[0m [2026-05-23 01:55:25,304 E 1461509 1461738] core_worker_process.cc:842: Failed to establish connection to the metrics exporter agent. Metrics will not be exported. Exporter agent status: RpcError: Running out of retries to initialize the metrics agent. rpc_code: 14[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122)
Loading safetensors checkpoint shards: 100% Completed | 4/4 [00:14<00:00, 3.50s/it][32m [repeated 23x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) The tokenizer you are loading from '/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the `fix_mistral_regex=True` flag when loading this tokenizer to fix this issue.[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m 2026-05-23 01:55:33.841 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1115 - Initializing OpenAIServingChat with custom_chat_template read from: chat_templates/qwen3_thinking_acc.jinja2[32m [repeated 13x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4876: UserWarning: barrier(): using the device under current context. You can specify `device_id` in `init_process_group` to mute this warning.
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m warnings.warn( # warn only once
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:38.114[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mload_checkpoints[0m:[36m1646[0m - [1m[32mSuccessfully loaded policy checkpoint[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:38.114[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36mload_checkpoints[0m:[36m1662[0m - [1m[32mSuccessfully loaded complete checkpoint state from global_step_18[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:38.114[0m | [1m[32mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_train_loop[0m:[36m455[0m - [1m[32mResumed training from global_step 18[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:38.132[0m | [1m[32mINFO [0m | [36mskyrl_train.utils.data_tracker[0m:[36mload_state[0m:[36m97[0m - [1m[32mLoaded data tracker state: epoch=0, consumed_in_epoch=1144, total_consumed=1144[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:38.133[0m | [33m[1mWARNING [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_train_loop[0m:[36m480[0m - [33m[1mData consumption count mismatch on resume: expected 1152, got 1144. This can happen after epoch boundary transitions or error recovery.[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:38.133[0m | [1m[32mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_train_loop[0m:[36m453[0m - [1m[32mFinished: 'load_checkpoints', time cost: 17.57s[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:38.133[0m | [1m[32mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_train_loop[0m:[36m487[0m - [1m[32mStarted: 'init_weight_sync_state'[0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m No module named 'vllm._version'
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m from .version import __version__, __version_tuple__ # isort:skip
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [32m2026-05-23 01:55:44.331[0m | [1m[32mINFO [0m | [36mlogging[0m:[36minfo[0m:[36m2216[0m - [1m[32m[weight-sync] Using master_addr=10.128.43.85, master_port=38311[0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m 2026-05-23 01:55:33.884 | INFO | skyrl_train.inference_engines.vllm.vllm_engine:_create_engine:1115 - Initializing OpenAIServingChat with custom_chat_template read from: chat_templates/qwen3_thinking_acc.jinja2[32m [repeated 3x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [32m2026-05-23 01:55:44.331[0m | [1m[32mINFO [0m | [36mlogging[0m:[36minfo[0m:[36m2216[0m - [1m[32m[weight-sync] get_node_ip_address()=10.128.43.85[0m[32m [repeated 3x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461509)[0m /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/__init__.py:7: RuntimeWarning: Failed to read commit hash:[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461509)[0m No module named 'vllm._version'[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=1461509)[0m from .version import __version__, __version_tuple__ # isort:skip[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank0]:[W523 01:55:44.016298483 socket.cpp:767] [c10d] The client socket cannot be initialized to connect to [jpbo-078-37-interconnect-1.jupiter.internal]:38311 (errno: 97 - Address family not supported by protocol).
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:44.431[0m | [1m[32mINFO [0m | [36mskyrl_train.fully_async_trainer[0m:[36m_train_loop[0m:[36m487[0m - [1m[32mFinished: 'init_weight_sync_state', time cost: 6.30s[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m --- Logging error in Loguru Handler #2 ---
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m Record was: {'elapsed': datetime.timedelta(seconds=226, microseconds=919551), 'exception': (type=<class 'ray.exceptions.RayTaskError(ActorDiedError)'>, value=RayTaskError(ActorDiedError)(ActorDiedError(RayTaskError('__init__', 'Traceback (most recent call last):\n File "python/ray/_raylet.pyx", line 1722, in ray._raylet.execute_task\n File "python/ray/_raylet.pyx", line 1659, in ray._raylet.execute_task.function_executor\n File "python/ray/_raylet.pyx", line 4342, in ray._raylet.CoreWorker.run_async_func_or_coro_in_event_loop\n File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 456, in result\n return self.__get_result()\n ^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result\n raise self._exception\n File "python/ray/_raylet.pyx", line 4329, in async_func\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/async_compat.py", line 52, in wrapper\n return func(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/function_manager.py", line 693, in actor_method_executor\n return method(__ray_actor, *args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/util/tracing/tracing_helper.py", line 461, in _resume_span\n return method(self, *_args, **_kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1028, in __init__\n super().__init__(*args, **kwargs)\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 500, in __init__\n self.llm = self._create_engine(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/util/tracing/tracing_helper.py", line 461, in _resume_span\n return method(self, *_args, **_kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1081, in _create_engine\n engine = vllm.AsyncLLMEngine.from_engine_args(engine_args, stat_loggers=stat_loggers)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 251, in from_engine_args\n return cls(\n ^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 148, in __init__\n self.engine_core = EngineCoreClient.make_async_mp_client(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 124, in make_async_mp_client\n return AsyncMPClient(*client_args)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 835, in __init__\n super().__init__(\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 490, in __init__\n with launch_core_engines(vllm_config, executor_class, log_stats) as (\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/contextlib.py", line 144, in __exit__\n next(self.gen)\n File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 936, in la
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m Traceback (most recent call last):
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/fully_async_trainer.py", line 434, in train
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m await self._train_loop()
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/fully_async_trainer.py", line 488, in _train_loop
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m self.init_weight_sync_state()
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/trainer.py", line 808, in init_weight_sync_state
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ray.get(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/auto_init_hook.py", line 22, in auto_init_wrapper
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return fn(*args, **kwargs)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/client_mode_hook.py", line 104, in wrapper
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return func(*args, **kwargs)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/worker.py", line 2967, in get
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m values, debugger_breakpoint = worker.get_objects(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/worker.py", line 1015, in get_objects
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m raise value.as_instanceof_cause()
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ray.exceptions.RayTaskError(ActorDiedError): [36mray::FSDPPolicyWorkerBase.init_weight_sync_state()[39m (pid=851205, ip=10.128.43.85, actor_id=8b87233f106bcf2285f12e1602000000, repr=<skyrl_train.workers.fsdp.fsdp_worker.FSDPPolicyWorkerBase object at 0x400dfe746e10>)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 449, in result
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return self.__get_result()
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m raise self._exception
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/workers/worker.py", line 299, in init_weight_sync_state
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m results = await asyncio.gather(*tasks)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/inference_engine_client.py", line 655, in init_weight_update_communicator
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m rank_offset_count += engine.tp_size() * engine.pp_size()
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/ray_wrapped_inference_engine.py", line 26, in tp_size
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return ray.get(self.inference_engine_actor.tp_size.remote())
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ray.exceptions.ActorDiedError: The actor died because of an error raised in its creation task, [36mray::AsyncVLLMInferenceEngine.__init__()[39m (pid=876507, ip=10.128.43.96, actor_id=58494295fdaf9ec8226b54f502000000, repr=<skyrl_train.inference_engines.vllm.vllm_engine.AsyncVLLMInferenceEngine object at 0x400be192c230>)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 456, in result
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return self.__get_result()
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m raise self._exception
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1028, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m super().__init__(*args, **kwargs)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 500, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m self.llm = self._create_engine(*args, **kwargs)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1081, in _create_engine
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m engine = vllm.AsyncLLMEngine.from_engine_args(engine_args, stat_loggers=stat_loggers)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 251, in from_engine_args
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return cls(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 148, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m self.engine_core = EngineCoreClient.make_async_mp_client(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 124, in make_async_mp_client
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return AsyncMPClient(*client_args)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 835, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m super().__init__(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 490, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m with launch_core_engines(vllm_config, executor_class, log_stats) as (
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/contextlib.py", line 144, in __exit__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m next(self.gen)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 936, in launch_core_engines
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m wait_for_engine_startup(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 995, in wait_for_engine_startup
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m raise RuntimeError(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m RuntimeError: Engine core initialization failed. See root cause above. Failed core proc(s): {}
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m During handling of the above exception, another exception occurred:
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m Traceback (most recent call last):
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/loguru/_handler.py", line 204, in emit
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m self._queue.put(str_record)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/multiprocessing/queues.py", line 393, in put
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m obj = _ForkingPickler.dumps(obj)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/multiprocessing/reduction.py", line 51, in dumps
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m cls(buf, protocol).dump(obj)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m _pickle.PicklingError: Can't pickle <class 'ray.exceptions.RayTaskError(ActorDiedError)'>: attribute lookup RayTaskError(ActorDiedError) on ray.exceptions failed
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m --- End of logging error ---
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py:470: UserWarning: No model update group to destroy
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) warnings.warn("No model update group to destroy")
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.513[0m | [1m[32mINFO [0m | [36mskyrl_train.inference_engines.inference_engine_client_http_endpoint[0m:[36mshutdown_server[0m:[36m203[0m - [1m[32mServer shut down after 2 seconds[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.513[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36m_guarded_sync[0m:[36m226[0m - [1m[32mHTTP endpoint shutdown complete[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.513[0m | [1m[32mINFO [0m | [36mexamples.terminal_bench.terminal_bench_generator[0m:[36mshutdown[0m:[36m328[0m - [1m[32mShutting down shared QueueOrchestrator...[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.516[0m | [1m[32mINFO [0m | [36mharbor.orchestrators.queue[0m:[36mshutdown[0m:[36m377[0m - [1m[32m[terminal_bench_generator:236] Shutdown complete. Total completed: 0[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.516[0m | [1m[32mINFO [0m | [36mexamples.terminal_bench.terminal_bench_generator[0m:[36mshutdown[0m:[36m330[0m - [1m[32mQueueOrchestrator shutdown complete[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.516[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36m_guarded_async[0m:[36m215[0m - [1m[32mGenerator shutdown complete[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.526[0m | [33m[1mWARNING [0m | [36mskyrl_train.trainer[0m:[36m_guarded_async[0m:[36m219[0m - [33m[1mInference engine teardown error (non-fatal): The actor died because of an error raised in its creation task, [36mray::AsyncVLLMInferenceEngine.__init__()[39m (pid=876507, ip=10.128.43.96, actor_id=58494295fdaf9ec8226b54f502000000, repr=<skyrl_train.inference_engines.vllm.vllm_engine.AsyncVLLMInferenceEngine object at 0x400be192c230>)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 456, in result
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return self.__get_result()
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m raise self._exception
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1028, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m super().__init__(*args, **kwargs)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 500, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m self.llm = self._create_engine(*args, **kwargs)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1081, in _create_engine
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m engine = vllm.AsyncLLMEngine.from_engine_args(engine_args, stat_loggers=stat_loggers)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 251, in from_engine_args
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return cls(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 148, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m self.engine_core = EngineCoreClient.make_async_mp_client(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 124, in make_async_mp_client
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m return AsyncMPClient(*client_args)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 835, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m super().__init__(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 490, in __init__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m with launch_core_engines(vllm_config, executor_class, log_stats) as (
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/contextlib.py", line 144, in __exit__
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m next(self.gen)
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 936, in launch_core_engines
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m wait_for_engine_startup(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 995, in wait_for_engine_startup
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m raise RuntimeError(
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m RuntimeError: Engine core initialization failed. See root cause above. Failed core proc(s): {}[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.526[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36m_kill_ray_actors[0m:[36m248[0m - [1m[32mKilling policy_model actors...[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.540[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36m_kill_ray_actors[0m:[36m267[0m - [1m[32mKilled 48 inference engine actor(s)[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.541[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36m_guarded_sync[0m:[36m226[0m - [1m[32mRay actor cleanup complete[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.542[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36m_kill_ray_actors[0m:[36m248[0m - [1m[32mKilling policy_model actors...[0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m [32m2026-05-23 01:55:45.544[0m | [1m[32mINFO [0m | [36mskyrl_train.trainer[0m:[36m_kill_ray_actors[0m:[36m267[0m - [1m[32mKilled 48 inference engine actor(s)[0m
|
|||
|
|
2026-05-23 01:55:45.559 | ERROR | __main__:main:125 - Training failed: [36mray::skyrl_entrypoint()[39m (pid=1461281, ip=10.128.43.81)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/examples/terminal_bench/entrypoints/main_tbench.py", line 96, in skyrl_entrypoint
|
|||
|
|
exp.run()
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/entrypoints/main_base.py", line 403, in run
|
|||
|
|
asyncio.run(trainer.train())
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/asyncio/runners.py", line 195, in run
|
|||
|
|
return runner.run(main)
|
|||
|
|
^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/asyncio/runners.py", line 118, in run
|
|||
|
|
return self._loop.run_until_complete(task)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/fully_async_trainer.py", line 434, in train
|
|||
|
|
await self._train_loop()
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/fully_async_trainer.py", line 488, in _train_loop
|
|||
|
|
self.init_weight_sync_state()
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/trainer.py", line 808, in init_weight_sync_state
|
|||
|
|
ray.get(
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
ray.exceptions.RayTaskError(ActorDiedError): [36mray::FSDPPolicyWorkerBase.init_weight_sync_state()[39m (pid=851205, ip=10.128.43.85, actor_id=8b87233f106bcf2285f12e1602000000, repr=<skyrl_train.workers.fsdp.fsdp_worker.FSDPPolicyWorkerBase object at 0x400dfe746e10>)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 449, in result
|
|||
|
|
return self.__get_result()
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
raise self._exception
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/workers/worker.py", line 299, in init_weight_sync_state
|
|||
|
|
results = await asyncio.gather(*tasks)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/inference_engine_client.py", line 655, in init_weight_update_communicator
|
|||
|
|
rank_offset_count += engine.tp_size() * engine.pp_size()
|
|||
|
|
^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/ray_wrapped_inference_engine.py", line 26, in tp_size
|
|||
|
|
return ray.get(self.inference_engine_actor.tp_size.remote())
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
ray.exceptions.ActorDiedError: The actor died because of an error raised in its creation task, [36mray::AsyncVLLMInferenceEngine.__init__()[39m (pid=876507, ip=10.128.43.96, actor_id=58494295fdaf9ec8226b54f502000000, repr=<skyrl_train.inference_engines.vllm.vllm_engine.AsyncVLLMInferenceEngine object at 0x400be192c230>)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 456, in result
|
|||
|
|
return self.__get_result()
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
raise self._exception
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1028, in __init__
|
|||
|
|
super().__init__(*args, **kwargs)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 500, in __init__
|
|||
|
|
self.llm = self._create_engine(*args, **kwargs)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1081, in _create_engine
|
|||
|
|
engine = vllm.AsyncLLMEngine.from_engine_args(engine_args, stat_loggers=stat_loggers)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 251, in from_engine_args
|
|||
|
|
return cls(
|
|||
|
|
^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 148, in __init__
|
|||
|
|
self.engine_core = EngineCoreClient.make_async_mp_client(
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 124, in make_async_mp_client
|
|||
|
|
return AsyncMPClient(*client_args)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 835, in __init__
|
|||
|
|
super().__init__(
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 490, in __init__
|
|||
|
|
with launch_core_engines(vllm_config, executor_class, log_stats) as (
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/contextlib.py", line 144, in __exit__
|
|||
|
|
next(self.gen)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 936, in launch_core_engines
|
|||
|
|
wait_for_engine_startup(
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 995, in wait_for_engine_startup
|
|||
|
|
raise RuntimeError(
|
|||
|
|
RuntimeError: Engine core initialization failed. See root cause above. Failed core proc(s): {}
|
|||
|
|
2026-05-23 01:55:45.559 | INFO | __main__:main:128 - Shutting down Ray on head node...
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) /e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py:470: UserWarning: No model update group to destroy[32m [repeated 21x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) warnings.warn("No model update group to destroy")[32m [repeated 21x across cluster][0m
|
|||
|
|
[36m(skyrl_entrypoint pid=1461281)[0m ⚙️ Running in WANDB offline mode
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m WARNING 05-23 01:54:39 [arg_utils.py:1256] The global random seed is set to 57. Since VLLM_ENABLE_V1_MULTIPROCESSING is set to False, this may affect the random state of the Python process that launched vLLM.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m INFO 05-23 01:54:39 [model.py:529] Resolved architecture: Qwen3ForCausalLM
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m INFO 05-23 01:54:39 [model.py:1549] Using max model len 32768
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m INFO 05-23 01:54:39 [arg_utils.py:1469] Using ray runtime env (env vars redacted): {'env_vars': {'NCCL_CUMEM_ENABLE': '***', 'RAY_ADDRESS': '***', 'VLLM_ALLOW_INSECURE_SERIALIZATION': '***', 'VLLM_ALLOW_RUNTIME_LORA_UPDATING': '***', 'VLLM_DISABLE_COMPILE_CACHE': '***', 'WANDB_API_KEY': '***'}}
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m INFO 05-23 01:54:39 [scheduler.py:224] Chunked prefill is enabled with max_num_batched_tokens=65536.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m INFO 05-23 01:54:39 [vllm.py:690] Asynchronous scheduling is enabled.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m WARNING 05-23 01:54:39 [vllm.py:728] Enforce eager set, overriding optimization level to -O0
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m INFO 05-23 01:54:39 [vllm.py:846] Cudagraph is disabled under eager mode
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m WARNING 05-23 01:54:39 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m WARNING 05-23 01:54:39 [system_utils.py:140] We must use the `spawn` multiprocessing start method. Overriding VLLM_WORKER_MULTIPROC_METHOD to 'spawn'. See https://docs.vllm.ai/en/latest/usage/troubleshooting.html#python-multiprocessing for more information. Reasons: In a Ray actor and can only be spawned
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m WARNING 05-23 01:54:44 [arg_utils.py:1256] The global random seed is set to 46. Since VLLM_ENABLE_V1_MULTIPROCESSING is set to False, this may affect the random state of the Python process that launched vLLM.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m INFO 05-23 01:54:44 [model.py:529] Resolved architecture: Qwen3ForCausalLM[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m INFO 05-23 01:54:44 [model.py:1549] Using max model len 32768[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m INFO 05-23 01:54:44 [arg_utils.py:1469] Using ray runtime env (env vars redacted): {'env_vars': {'NCCL_CUMEM_ENABLE': '***', 'RAY_ADDRESS': '***', 'VLLM_ALLOW_INSECURE_SERIALIZATION': '***', 'VLLM_ALLOW_RUNTIME_LORA_UPDATING': '***', 'VLLM_DISABLE_COMPILE_CACHE': '***', 'WANDB_API_KEY': '***'}}[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m INFO 05-23 01:54:44 [scheduler.py:224] Chunked prefill is enabled with max_num_batched_tokens=65536.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m INFO 05-23 01:54:44 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m WARNING 05-23 01:54:44 [vllm.py:728] Enforce eager set, overriding optimization level to -O0[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m INFO 05-23 01:54:44 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m WARNING 05-23 01:54:44 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m WARNING 05-23 01:54:44 [system_utils.py:140] We must use the `spawn` multiprocessing start method. Overriding VLLM_WORKER_MULTIPROC_METHOD to 'spawn'. See https://docs.vllm.ai/en/latest/usage/troubleshooting.html#python-multiprocessing for more information. Reasons: In a Ray actor and can only be spawned[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874281) INFO 05-23 01:54:48 [core.py:97] Initializing a V1 LLM engine (vdev) with config: model='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', speculative_config=None, tokenizer='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=32768, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=True, enable_return_routed_experts=False, kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, kv_cache_metrics=False, kv_cache_metrics_sample=0.01, cudagraph_metrics=False, enable_layerwise_nvtx_tracing=False, enable_mfu_metrics=False, enable_mm_processor_stats=False, enable_logging_iteration_details=False), seed=57, served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'level': None, 'mode': <CompilationMode.NONE: 0>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inductor', 'custom_ops': ['all'], 'splitting_ops': [], 'compile_mm_encoder': False, 'compile_sizes': [], 'compile_ranges_split_points': [65536], 'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_passes': {}, 'cudagraph_mode': <CUDAGraphMode.NONE: 0>, 'cudagraph_num_of_warmups': 0, 'cudagraph_capture_sizes': [], 'cudagraph_copy_inputs': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {'fuse_norm_quant': False, 'fuse_act_quant': False, 'fuse_attn_quant': False, 'eliminate_noops': False, 'enable_sp': False, 'fuse_gemm_comms': False, 'fuse_allreduce_rms': False, 'fuse_act_padding': False}, 'max_cudagraph_capture_size': 0, 'dynamic_shapes_config': {'type': <DynamicShapesType.BACKED: 'backed'>, 'evaluate_guards': False, 'assume_32_bit_indexing': False}, 'local_cache_dir': None, 'fast_moe_cold_start': True, 'static_all_moe_layers': []}
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m WARNING 05-23 01:54:49 [arg_utils.py:1256] The global random seed is set to 80. Since VLLM_ENABLE_V1_MULTIPROCESSING is set to False, this may affect the random state of the Python process that launched vLLM.[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m INFO 05-23 01:54:49 [model.py:529] Resolved architecture: Qwen3ForCausalLM[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m INFO 05-23 01:54:49 [model.py:1549] Using max model len 32768[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m INFO 05-23 01:54:49 [arg_utils.py:1469] Using ray runtime env (env vars redacted): {'env_vars': {'NCCL_CUMEM_ENABLE': '***', 'RAY_ADDRESS': '***', 'VLLM_ALLOW_INSECURE_SERIALIZATION': '***', 'VLLM_ALLOW_RUNTIME_LORA_UPDATING': '***', 'VLLM_DISABLE_COMPILE_CACHE': '***', 'WANDB_API_KEY': '***'}}[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m INFO 05-23 01:54:49 [scheduler.py:224] Chunked prefill is enabled with max_num_batched_tokens=65536.[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m INFO 05-23 01:54:49 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m WARNING 05-23 01:54:49 [vllm.py:728] Enforce eager set, overriding optimization level to -O0[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m INFO 05-23 01:54:49 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m WARNING 05-23 01:54:49 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900914, ip=10.128.43.91)[0m WARNING 05-23 01:54:49 [system_utils.py:140] We must use the `spawn` multiprocessing start method. Overriding VLLM_WORKER_MULTIPROC_METHOD to 'spawn'. See https://docs.vllm.ai/en/latest/usage/troubleshooting.html#python-multiprocessing for more information. Reasons: In a Ray actor and can only be spawned[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=874053, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874281) INFO 05-23 01:54:52 [worker_base.py:289] Injected <class 'skyrl_train.inference_engines.vllm.vllm_engine.WorkerWrap'> into <class 'vllm.v1.worker.gpu_worker.Worker'> for extended collective_rpc calls ['_apply_fp8_weight_loader_patches', '_is_fp8_model', '_quantize_weights_for_fp8', '_restore_param_subclasses', '_undo_param_subclasses', 'begin_weight_update', 'destroy_weights_update_group', 'end_weight_update', 'init_weight_update_communicator', 'load_weights', 'set_numa_affinity', 'test_rpc']
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=838979, ip=10.128.43.94)[0m (EngineCore_DP0 pid=839337) INFO 05-23 01:54:53 [core.py:97] Initializing a V1 LLM engine (vdev) with config: model='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', speculative_config=None, tokenizer='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=32768, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=True, enable_return_routed_experts=False, kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, kv_cache_metrics=False, kv_cache_metrics_sample=0.01, cudagraph_metrics=False, enable_layerwise_nvtx_tracing=False, enable_mfu_metrics=False, enable_mm_processor_stats=False, enable_logging_iteration_details=False), seed=46, served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'level': None, 'mode': <CompilationMode.NONE: 0>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inductor', 'custom_ops': ['all'], 'splitting_ops': [], 'compile_mm_encoder': False, 'compile_sizes': [], 'compile_ranges_split_points': [65536], 'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_passes': {}, 'cudagraph_mode': <CUDAGraphMode.NONE: 0>, 'cudagraph_num_of_warmups': 0, 'cudagraph_capture_sizes': [], 'cudagraph_copy_inputs': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {'fuse_norm_quant': False, 'fuse_act_quant': False, 'fuse_attn_quant': False, 'eliminate_noops': False, 'enable_sp': False, 'fuse_gemm_comms': False, 'fuse_allreduce_rms': False, 'fuse_act_padding': False}, 'max_cudagraph_capture_size': 0, 'dynamic_shapes_config': {'type': <DynamicShapesType.BACKED: 'backed'>, 'evaluate_guards': False, 'assume_32_bit_indexing': False}, 'local_cache_dir': None, 'fast_moe_cold_start': True, 'static_all_moe_layers': []}[32m [repeated 7x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307) INFO 05-23 01:54:54 [parallel_state.py:1234] world_size=1 rank=0 local_rank=0 distributed_init_method=tcp://10.128.43.93:49029 backend=nccl
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307) INFO 05-23 01:54:54 [parallel_state.py:1445] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, PCP rank 0, TP rank 0, EP rank N/A
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m WARNING 05-23 01:54:54 [arg_utils.py:1256] The global random seed is set to 68. Since VLLM_ENABLE_V1_MULTIPROCESSING is set to False, this may affect the random state of the Python process that launched vLLM.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m INFO 05-23 01:54:54 [model.py:529] Resolved architecture: Qwen3ForCausalLM[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m INFO 05-23 01:54:54 [model.py:1549] Using max model len 32768[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m INFO 05-23 01:54:54 [arg_utils.py:1469] Using ray runtime env (env vars redacted): {'env_vars': {'NCCL_CUMEM_ENABLE': '***', 'RAY_ADDRESS': '***', 'VLLM_ALLOW_INSECURE_SERIALIZATION': '***', 'VLLM_ALLOW_RUNTIME_LORA_UPDATING': '***', 'VLLM_DISABLE_COMPILE_CACHE': '***', 'WANDB_API_KEY': '***'}}[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m INFO 05-23 01:54:54 [scheduler.py:224] Chunked prefill is enabled with max_num_batched_tokens=65536.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m INFO 05-23 01:54:54 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m WARNING 05-23 01:54:54 [vllm.py:728] Enforce eager set, overriding optimization level to -O0[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m INFO 05-23 01:54:54 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(pid=851205, ip=10.128.43.85)[0m ⚙️ Running in WANDB offline mode
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307) INFO 05-23 01:54:54 [gpu_model_runner.py:4125] Starting to load model /e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6...
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m WARNING 05-23 01:54:54 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m WARNING 05-23 01:54:54 [system_utils.py:140] We must use the `spawn` multiprocessing start method. Overriding VLLM_WORKER_MULTIPROC_METHOD to 'spawn'. See https://docs.vllm.ai/en/latest/usage/troubleshooting.html#python-multiprocessing for more information. Reasons: In a Ray actor and can only be spawned[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=873920, ip=10.128.43.93)[0m (EngineCore_DP0 pid=874307) INFO 05-23 01:54:56 [cuda.py:367] Using FLASH_ATTN attention backend out of potential backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION'].
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) INFO 05-23 01:54:57 [worker_base.py:289] Injected <class 'skyrl_train.inference_engines.vllm.vllm_engine.WorkerWrap'> into <class 'vllm.v1.worker.gpu_worker.Worker'> for extended collective_rpc calls ['_apply_fp8_weight_loader_patches', '_is_fp8_model', '_quantize_weights_for_fp8', '_restore_param_subclasses', '_undo_param_subclasses', 'begin_weight_update', 'destroy_weights_update_group', 'end_weight_update', 'init_weight_update_communicator', 'load_weights', 'set_numa_affinity', 'test_rpc'][32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900916, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901154) INFO 05-23 01:54:58 [core.py:97] Initializing a V1 LLM engine (vdev) with config: model='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', speculative_config=None, tokenizer='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=32768, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=True, enable_return_routed_experts=False, kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, kv_cache_metrics=False, kv_cache_metrics_sample=0.01, cudagraph_metrics=False, enable_layerwise_nvtx_tracing=False, enable_mfu_metrics=False, enable_mm_processor_stats=False, enable_logging_iteration_details=False), seed=79, served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'level': None, 'mode': <CompilationMode.NONE: 0>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inductor', 'custom_ops': ['all'], 'splitting_ops': [], 'compile_mm_encoder': False, 'compile_sizes': [], 'compile_ranges_split_points': [65536], 'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_passes': {}, 'cudagraph_mode': <CUDAGraphMode.NONE: 0>, 'cudagraph_num_of_warmups': 0, 'cudagraph_capture_sizes': [], 'cudagraph_copy_inputs': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {'fuse_norm_quant': False, 'fuse_act_quant': False, 'fuse_attn_quant': False, 'eliminate_noops': False, 'enable_sp': False, 'fuse_gemm_comms': False, 'fuse_allreduce_rms': False, 'fuse_act_padding': False}, 'max_cudagraph_capture_size': 0, 'dynamic_shapes_config': {'type': <DynamicShapesType.BACKED: 'backed'>, 'evaluate_guards': False, 'assume_32_bit_indexing': False}, 'local_cache_dir': None, 'fast_moe_cold_start': True, 'static_all_moe_layers': []}[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] EngineCore failed to start.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] Traceback (most recent call last):
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 996, in run_engine_core
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] engine_core = EngineCoreProc(*args, engine_index=dp_rank, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 740, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] super().__init__(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 106, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] self.model_executor = executor_class(vllm_config)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/executor/abstract.py", line 103, in __init__
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] self._init_executor()
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/executor/uniproc_executor.py", line 47, in _init_executor
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] self.driver_worker.init_device()
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/worker/worker_base.py", line 332, in init_device
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] self.worker.init_device() # type: ignore
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/worker/gpu_worker.py", line 232, in init_device
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] init_worker_distributed_environment(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/worker/gpu_worker.py", line 1040, in init_worker_distributed_environment
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] init_distributed_environment(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/distributed/parallel_state.py", line 1256, in init_distributed_environment
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] torch.distributed.init_process_group(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] return func(*args, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/c10d_logger.py", line 95, in wrapper
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] func_return = func(*args, **kwargs)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py", line 1762, in init_process_group
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] store, rank, world_size = next(rendezvous_iterator)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/rendezvous.py", line 230, in _tcp_rendezvous_handler
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] store = _create_c10d_store(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^^^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/distributed/rendezvous.py", line 198, in _create_c10d_store
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] return TCPStore(
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] ^^^^^^^^^
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] torch.distributed.DistNetworkError: The server socket has failed to listen on any local network address. port: 36099, useIpv6: false, code: -98, name: EADDRINUSE, message: address already in use
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] Exception raised from makeWithPort at /pytorch/torch/csrc/distributed/c10d/TCPStoreLibUvBackend.cpp:307 (most recent call first):
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) + 0xb0 (0x40004b8fc700 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libc10.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #1: <unknown function> + 0x5f29220 (0x40002ba99220 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #2: <unknown function> + 0x5f4326c (0x40002bab326c in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #3: <unknown function> + 0x5f49074 (0x40002bab9074 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #4: <unknown function> + 0x5f49138 (0x40002bab9138 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #5: <unknown function> + 0x5f2ccc4 (0x40002ba9ccc4 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #6: c10d::TCPStore::TCPStore(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, c10d::TCPStoreOptions const&) + 0x104 (0x40002baa1564 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_cpu.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #7: <unknown function> + 0x109a094 (0x40002584a094 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #8: <unknown function> + 0x113236c (0x4000258e236c in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #9: <unknown function> + 0x5d6d60 (0x400024d86d60 in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #10: <unknown function> + 0x1b7a38 (0xaaaab65b7a38 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #11: _PyObject_MakeTpCall + 0x98 (0xaaaab6565db8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #12: <unknown function> + 0x169f50 (0xaaaab6569f50 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #13: <unknown function> + 0x1682e4 (0xaaaab65682e4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #14: <unknown function> + 0x1e0ce8 (0xaaaab65e0ce8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #15: <unknown function> + 0x1d7ddc (0xaaaab65d7ddc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #16: <unknown function> + 0x646b0c (0x400024df6b0c in /e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/torch/lib/libtorch_python.so)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #17: _PyObject_MakeTpCall + 0x98 (0xaaaab6565db8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #18: _PyEval_EvalFrameDefault + 0x280c (0xaaaab666ae54 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #19: <unknown function> + 0x1808c0 (0xaaaab65808c0 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #20: <unknown function> + 0x182bf8 (0xaaaab6582bf8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #21: <unknown function> + 0x25fd30 (0xaaaab665fd30 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #22: <unknown function> + 0x1b7d20 (0xaaaab65b7d20 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #23: PyObject_Vectorcall + 0x54 (0xaaaab65660e4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #24: _PyEval_EvalFrameDefault + 0x280c (0xaaaab666ae54 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #25: _PyObject_FastCallDictTstate + 0x80 (0xaaaab6567f20 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #26: _PyObject_Call_Prepend + 0x140 (0xaaaab65681ec in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #27: <unknown function> + 0x1e0d80 (0xaaaab65e0d80 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #28: <unknown function> + 0x1d7ddc (0xaaaab65d7ddc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #29: _PyObject_MakeTpCall + 0x98 (0xaaaab6565db8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #30: _PyEval_EvalFrameDefault + 0x280c (0xaaaab666ae54 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #31: _PyObject_FastCallDictTstate + 0x10c (0xaaaab6567fac in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #32: _PyObject_Call_Prepend + 0x140 (0xaaaab65681ec in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #33: <unknown function> + 0x1e0d80 (0xaaaab65e0d80 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #34: <unknown function> + 0x1d7ddc (0xaaaab65d7ddc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #35: _PyObject_Call + 0x68 (0xaaaab6568488 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #36: _PyEval_EvalFrameDefault + 0x52ac (0xaaaab666d8f4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #37: PyEval_EvalCode + 0xb4 (0xaaaab6672eb4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #38: <unknown function> + 0x2ccdcc (0xaaaab66ccdcc in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #39: <unknown function> + 0x2ccef4 (0xaaaab66ccef4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #40: PyRun_StringFlags + 0x90 (0xaaaab66d1050 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #41: PyRun_SimpleStringFlags + 0x58 (0xaaaab66d10f8 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #42: Py_RunMain + 0x2c8 (0xaaaab66f9190 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #43: Py_BytesMain + 0x64 (0xaaaab66f9fb4 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #44: <unknown function> + 0x27540 (0x400013df7540 in /lib64/libc.so.6)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #45: __libc_start_main + 0x98 (0x400013df7618 in /lib64/libc.so.6)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006] frame #46: <unknown function> + 0x10e0c0 (0xaaaab650e0c0 in VLLM::EngineCore)
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) ERROR 05-23 01:54:59 [core.py:1006]
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876507, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876748) INFO 05-23 01:54:59 [parallel_state.py:1234] world_size=1 rank=0 local_rank=0 distributed_init_method=tcp://10.128.43.96:36099 backend=nccl[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876509, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876761) INFO 05-23 01:54:59 [parallel_state.py:1445] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, PCP rank 0, TP rank 0, EP rank N/A[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m WARNING 05-23 01:54:59 [arg_utils.py:1256] The global random seed is set to 89. Since VLLM_ENABLE_V1_MULTIPROCESSING is set to False, this may affect the random state of the Python process that launched vLLM.[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m INFO 05-23 01:54:59 [model.py:529] Resolved architecture: Qwen3ForCausalLM[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m INFO 05-23 01:54:59 [model.py:1549] Using max model len 32768[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m INFO 05-23 01:54:59 [arg_utils.py:1469] Using ray runtime env (env vars redacted): {'env_vars': {'NCCL_CUMEM_ENABLE': '***', 'RAY_ADDRESS': '***', 'VLLM_ALLOW_INSECURE_SERIALIZATION': '***', 'VLLM_ALLOW_RUNTIME_LORA_UPDATING': '***', 'VLLM_DISABLE_COMPILE_CACHE': '***', 'WANDB_API_KEY': '***'}}[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m INFO 05-23 01:54:59 [scheduler.py:224] Chunked prefill is enabled with max_num_batched_tokens=65536.[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m INFO 05-23 01:54:59 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m WARNING 05-23 01:54:59 [vllm.py:728] Enforce eager set, overriding optimization level to -O0[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m INFO 05-23 01:54:59 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876509, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876761) INFO 05-23 01:54:59 [gpu_model_runner.py:4125] Starting to load model /e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6...[32m [repeated 17x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m WARNING 05-23 01:54:59 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m WARNING 05-23 01:54:59 [system_utils.py:140] We must use the `spawn` multiprocessing start method. Overriding VLLM_WORKER_MULTIPROC_METHOD to 'spawn'. See https://docs.vllm.ai/en/latest/usage/troubleshooting.html#python-multiprocessing for more information. Reasons: In a Ray actor and can only be spawned[32m [repeated 10x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:00 [cuda.py:367] Using FLASH_ATTN attention backend out of potential backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION'].[32m [repeated 18x across cluster][0m
|
|||
|
|
[36m(pid=851281, ip=10.128.43.85)[0m ⚙️ Running in WANDB offline mode
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003033, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003273) INFO 05-23 01:55:02 [worker_base.py:289] Injected <class 'skyrl_train.inference_engines.vllm.vllm_engine.WorkerWrap'> into <class 'vllm.v1.worker.gpu_worker.Worker'> for extended collective_rpc calls ['_apply_fp8_weight_loader_patches', '_is_fp8_model', '_quantize_weights_for_fp8', '_restore_param_subclasses', '_undo_param_subclasses', 'begin_weight_update', 'destroy_weights_update_group', 'end_weight_update', 'init_weight_update_communicator', 'load_weights', 'set_numa_affinity', 'test_rpc'][32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003034, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003287) INFO 05-23 01:55:00 [core.py:97] Initializing a V1 LLM engine (vdev) with config: model='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', speculative_config=None, tokenizer='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=32768, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=True, enable_return_routed_experts=False, kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, kv_cache_metrics=False, kv_cache_metrics_sample=0.01, cudagraph_metrics=False, enable_layerwise_nvtx_tracing=False, enable_mfu_metrics=False, enable_mm_processor_stats=False, enable_logging_iteration_details=False), seed=64, served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'level': None, 'mode': <CompilationMode.NONE: 0>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inductor', 'custom_ops': ['all'], 'splitting_ops': [], 'compile_mm_encoder': False, 'compile_sizes': [], 'compile_ranges_split_points': [65536], 'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_passes': {}, 'cudagraph_mode': <CUDAGraphMode.NONE: 0>, 'cudagraph_num_of_warmups': 0, 'cudagraph_capture_sizes': [], 'cudagraph_copy_inputs': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {'fuse_norm_quant': False, 'fuse_act_quant': False, 'fuse_attn_quant': False, 'eliminate_noops': False, 'enable_sp': False, 'fuse_gemm_comms': False, 'fuse_allreduce_rms': False, 'fuse_act_padding': False}, 'max_cudagraph_capture_size': 0, 'dynamic_shapes_config': {'type': <DynamicShapesType.BACKED: 'backed'>, 'evaluate_guards': False, 'assume_32_bit_indexing': False}, 'local_cache_dir': None, 'fast_moe_cold_start': True, 'static_all_moe_layers': []}[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1003034, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003287) INFO 05-23 01:55:03 [parallel_state.py:1234] world_size=1 rank=0 local_rank=0 distributed_init_method=tcp://10.128.43.108:56203 backend=nccl[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1002904, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003263) INFO 05-23 01:55:04 [parallel_state.py:1445] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, PCP rank 0, TP rank 0, EP rank N/A[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m WARNING 05-23 01:55:00 [arg_utils.py:1256] The global random seed is set to 84. Since VLLM_ENABLE_V1_MULTIPROCESSING is set to False, this may affect the random state of the Python process that launched vLLM.[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m INFO 05-23 01:55:00 [model.py:529] Resolved architecture: Qwen3ForCausalLM[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m INFO 05-23 01:55:00 [model.py:1549] Using max model len 32768[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m INFO 05-23 01:55:00 [arg_utils.py:1469] Using ray runtime env (env vars redacted): {'env_vars': {'NCCL_CUMEM_ENABLE': '***', 'RAY_ADDRESS': '***', 'VLLM_ALLOW_INSECURE_SERIALIZATION': '***', 'VLLM_ALLOW_RUNTIME_LORA_UPDATING': '***', 'VLLM_DISABLE_COMPILE_CACHE': '***', 'WANDB_API_KEY': '***'}}[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m INFO 05-23 01:55:00 [scheduler.py:224] Chunked prefill is enabled with max_num_batched_tokens=65536.[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m INFO 05-23 01:55:00 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m WARNING 05-23 01:55:00 [vllm.py:728] Enforce eager set, overriding optimization level to -O0[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m INFO 05-23 01:55:00 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1002904, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003263) INFO 05-23 01:55:04 [gpu_model_runner.py:4125] Starting to load model /e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6...[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m WARNING 05-23 01:55:00 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m WARNING 05-23 01:55:00 [system_utils.py:140] We must use the `spawn` multiprocessing start method. Overriding VLLM_WORKER_MULTIPROC_METHOD to 'spawn'. See https://docs.vllm.ai/en/latest/usage/troubleshooting.html#python-multiprocessing for more information. Reasons: In a Ray actor and can only be spawned[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1002904, ip=10.128.43.108)[0m (EngineCore_DP0 pid=1003263) INFO 05-23 01:55:05 [cuda.py:367] Using FLASH_ATTN attention backend out of potential backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION'].[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(pid=1461509)[0m ⚙️ Running in WANDB offline mode[32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095514, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095768) INFO 05-23 01:55:07 [worker_base.py:289] Injected <class 'skyrl_train.inference_engines.vllm.vllm_engine.WorkerWrap'> into <class 'vllm.v1.worker.gpu_worker.Worker'> for extended collective_rpc calls ['_apply_fp8_weight_loader_patches', '_is_fp8_model', '_quantize_weights_for_fp8', '_restore_param_subclasses', '_undo_param_subclasses', 'begin_weight_update', 'destroy_weights_update_group', 'end_weight_update', 'init_weight_update_communicator', 'load_weights', 'set_numa_affinity', 'test_rpc'][32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117) INFO 05-23 01:55:08 [core.py:97] Initializing a V1 LLM engine (vdev) with config: model='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', speculative_config=None, tokenizer='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=32768, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=True, enable_return_routed_experts=False, kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, kv_cache_metrics=False, kv_cache_metrics_sample=0.01, cudagraph_metrics=False, enable_layerwise_nvtx_tracing=False, enable_mfu_metrics=False, enable_mm_processor_stats=False, enable_logging_iteration_details=False), seed=86, served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'level': None, 'mode': <CompilationMode.NONE: 0>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inductor', 'custom_ops': ['all'], 'splitting_ops': [], 'compile_mm_encoder': False, 'compile_sizes': [], 'compile_ranges_split_points': [65536], 'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_passes': {}, 'cudagraph_mode': <CUDAGraphMode.NONE: 0>, 'cudagraph_num_of_warmups': 0, 'cudagraph_capture_sizes': [], 'cudagraph_copy_inputs': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {'fuse_norm_quant': False, 'fuse_act_quant': False, 'fuse_attn_quant': False, 'eliminate_noops': False, 'enable_sp': False, 'fuse_gemm_comms': False, 'fuse_allreduce_rms': False, 'fuse_act_padding': False}, 'max_cudagraph_capture_size': 0, 'dynamic_shapes_config': {'type': <DynamicShapesType.BACKED: 'backed'>, 'evaluate_guards': False, 'assume_32_bit_indexing': False}, 'local_cache_dir': None, 'fast_moe_cold_start': True, 'static_all_moe_layers': []}[32m [repeated 15x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095642, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095747) INFO 05-23 01:55:08 [parallel_state.py:1234] world_size=1 rank=0 local_rank=0 distributed_init_method=tcp://10.128.43.86:50391 backend=nccl[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:09 [default_loader.py:293] Loading weights took 8.80 seconds
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095751) INFO 05-23 01:55:09 [parallel_state.py:1445] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, PCP rank 0, TP rank 0, EP rank N/A[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:09 [gpu_model_runner.py:4222] Model loading took 15.27 GiB memory and 9.163299 seconds
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2095386, ip=10.128.43.86)[0m (EngineCore_DP0 pid=2095751) INFO 05-23 01:55:09 [gpu_model_runner.py:4125] Starting to load model /e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6...[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552034, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552274) INFO 05-23 01:55:11 [cuda.py:367] Using FLASH_ATTN attention backend out of potential backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION'].[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:12 [gpu_worker.py:373] Available KV cache memory: 49.32 GiB
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:12 [kv_cache_utils.py:1307] GPU KV cache size: 359,104 tokens
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:12 [kv_cache_utils.py:1312] Maximum concurrency for 32,768 tokens per request: 10.96x
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868952, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869183) INFO 05-23 01:55:11 [worker_base.py:289] Injected <class 'skyrl_train.inference_engines.vllm.vllm_engine.WorkerWrap'> into <class 'vllm.v1.worker.gpu_worker.Worker'> for extended collective_rpc calls ['_apply_fp8_weight_loader_patches', '_is_fp8_model', '_quantize_weights_for_fp8', '_restore_param_subclasses', '_undo_param_subclasses', 'begin_weight_update', 'destroy_weights_update_group', 'end_weight_update', 'init_weight_update_communicator', 'load_weights', 'set_numa_affinity', 'test_rpc'][32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:12 [kernel_warmup.py:44] Skipping FlashInfer autotune because it is disabled.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:12 [core.py:278] init engine (profile, create kv cache, warmup model) took 3.11 seconds
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) WARNING 05-23 01:55:13 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:13 [vllm.py:690] Asynchronous scheduling is enabled.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) WARNING 05-23 01:55:13 [vllm.py:735] Inductor compilation was disabled by user settings, optimizations settings that are only active during inductor compilation will be ignored.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m (EngineCore_DP0 pid=876753) INFO 05-23 01:55:13 [vllm.py:846] Cudagraph is disabled under eager mode
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=876379, ip=10.128.43.96)[0m WARNING 05-23 01:55:13 [model.py:1350] Default vLLM sampling parameters have been overridden by the model's `generation_config.json`: `{'temperature': 0.6, 'top_k': 20, 'top_p': 0.95}`. If this is not intended, please relaunch vLLM instance with `--generation-config vllm`.
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=868953, ip=10.128.43.90)[0m (EngineCore_DP0 pid=869207) INFO 05-23 01:55:09 [core.py:97] Initializing a V1 LLM engine (vdev) with config: model='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', speculative_config=None, tokenizer='/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=32768, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=True, enable_return_routed_experts=False, kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, kv_cache_metrics=False, kv_cache_metrics_sample=0.01, cudagraph_metrics=False, enable_layerwise_nvtx_tracing=False, enable_mfu_metrics=False, enable_mm_processor_stats=False, enable_logging_iteration_details=False), seed=84, served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'level': None, 'mode': <CompilationMode.NONE: 0>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inductor', 'custom_ops': ['all'], 'splitting_ops': [], 'compile_mm_encoder': False, 'compile_sizes': [], 'compile_ranges_split_points': [65536], 'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_passes': {}, 'cudagraph_mode': <CUDAGraphMode.NONE: 0>, 'cudagraph_num_of_warmups': 0, 'cudagraph_capture_sizes': [], 'cudagraph_copy_inputs': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {'fuse_norm_quant': False, 'fuse_act_quant': False, 'fuse_attn_quant': False, 'eliminate_noops': False, 'enable_sp': False, 'fuse_gemm_comms': False, 'fuse_allreduce_rms': False, 'fuse_act_padding': False}, 'max_cudagraph_capture_size': 0, 'dynamic_shapes_config': {'type': <DynamicShapesType.BACKED: 'backed'>, 'evaluate_guards': False, 'assume_32_bit_indexing': False}, 'local_cache_dir': None, 'fast_moe_cold_start': True, 'static_all_moe_layers': []}[32m [repeated 5x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) INFO 05-23 01:55:14 [parallel_state.py:1234] world_size=1 rank=0 local_rank=0 distributed_init_method=tcp://10.128.43.103:43683 backend=nccl[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) INFO 05-23 01:55:14 [default_loader.py:293] Loading weights took 13.88 seconds[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117) INFO 05-23 01:55:14 [parallel_state.py:1445] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, PCP rank 0, TP rank 0, EP rank N/A[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) INFO 05-23 01:55:14 [gpu_model_runner.py:4222] Model loading took 15.27 GiB memory and 14.858314 seconds[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117) INFO 05-23 01:55:14 [gpu_model_runner.py:4125] Starting to load model /e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6...[32m [repeated 16x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) INFO 05-23 01:55:15 [cuda.py:367] Using FLASH_ATTN attention backend out of potential backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION'].[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658393) INFO 05-23 01:55:17 [gpu_worker.py:373] Available KV cache memory: 49.32 GiB[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658393) INFO 05-23 01:55:17 [kv_cache_utils.py:1307] GPU KV cache size: 359,104 tokens[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658393) INFO 05-23 01:55:17 [kv_cache_utils.py:1312] Maximum concurrency for 32,768 tokens per request: 10.96x[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117) INFO 05-23 01:55:14 [worker_base.py:289] Injected <class 'skyrl_train.inference_engines.vllm.vllm_engine.WorkerWrap'> into <class 'vllm.v1.worker.gpu_worker.Worker'> for extended collective_rpc calls ['_apply_fp8_weight_loader_patches', '_is_fp8_model', '_quantize_weights_for_fp8', '_restore_param_subclasses', '_undo_param_subclasses', 'begin_weight_update', 'destroy_weights_update_group', 'end_weight_update', 'init_weight_update_communicator', 'load_weights', 'set_numa_affinity', 'test_rpc'][32m [repeated 6x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658393) INFO 05-23 01:55:17 [kernel_warmup.py:44] Skipping FlashInfer autotune because it is disabled.[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) INFO 05-23 01:55:17 [core.py:278] init engine (profile, create kv cache, warmup model) took 2.76 seconds[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) WARNING 05-23 01:55:17 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) INFO 05-23 01:55:17 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) WARNING 05-23 01:55:17 [vllm.py:735] Inductor compilation was disabled by user settings, optimizations settings that are only active during inductor compilation will be ignored.[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658162, ip=10.128.43.87)[0m (EngineCore_DP0 pid=1658405) INFO 05-23 01:55:17 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m NCCL version 2.27.7+cuda13.0
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=1658163, ip=10.128.43.87)[0m WARNING 05-23 01:55:18 [model.py:1350] Default vLLM sampling parameters have been overridden by the model's `generation_config.json`: `{'temperature': 0.6, 'top_k': 20, 'top_p': 0.95}`. If this is not intended, please relaunch vLLM instance with `--generation-config vllm`.[32m [repeated 22x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975760, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976117) INFO 05-23 01:55:14 [parallel_state.py:1234] world_size=1 rank=0 local_rank=0 distributed_init_method=tcp://10.128.43.103:58483 backend=nccl
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900915, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901149) INFO 05-23 01:55:17 [default_loader.py:293] Loading weights took 13.60 seconds[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900915, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901149) INFO 05-23 01:55:18 [gpu_model_runner.py:4222] Model loading took 15.27 GiB memory and 14.850248 seconds[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank-0]: Loading model from /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints/global_step_18/policy/model_world_size_8_rank_0.pt
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank-0]: Loading extra_state from /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints/global_step_18/policy/extra_state_world_size_8_rank_0.pt
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank-0]: Loading optim from /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints/global_step_18/policy/optim_world_size_8_rank_0.pt
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) INFO 05-23 01:55:20 [gpu_worker.py:373] Available KV cache memory: 49.32 GiB[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) INFO 05-23 01:55:20 [kv_cache_utils.py:1307] GPU KV cache size: 359,104 tokens[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) INFO 05-23 01:55:20 [kv_cache_utils.py:1312] Maximum concurrency for 32,768 tokens per request: 10.96x[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900915, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901149) INFO 05-23 01:55:21 [kernel_warmup.py:44] Skipping FlashInfer autotune because it is disabled.[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) INFO 05-23 01:55:21 [core.py:278] init engine (profile, create kv cache, warmup model) took 2.81 seconds[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) WARNING 05-23 01:55:21 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) INFO 05-23 01:55:21 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) WARNING 05-23 01:55:21 [vllm.py:735] Inductor compilation was disabled by user settings, optimizations settings that are only active during inductor compilation will be ignored.[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m (EngineCore_DP0 pid=901141) INFO 05-23 01:55:21 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=900781, ip=10.128.43.91)[0m WARNING 05-23 01:55:21 [model.py:1350] Default vLLM sampling parameters have been overridden by the model's `generation_config.json`: `{'temperature': 0.6, 'top_k': 20, 'top_p': 0.95}`. If this is not intended, please relaunch vLLM instance with `--generation-config vllm`.[32m [repeated 4x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940164, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940526) INFO 05-23 01:55:25 [default_loader.py:293] Loading weights took 13.84 seconds[32m [repeated 9x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267) INFO 05-23 01:55:25 [gpu_model_runner.py:4222] Model loading took 15.27 GiB memory and 14.506031 seconds[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531) INFO 05-23 01:55:28 [gpu_worker.py:373] Available KV cache memory: 49.32 GiB[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531) INFO 05-23 01:55:28 [kv_cache_utils.py:1307] GPU KV cache size: 359,104 tokens[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531) INFO 05-23 01:55:28 [kv_cache_utils.py:1312] Maximum concurrency for 32,768 tokens per request: 10.96x[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531) INFO 05-23 01:55:29 [kernel_warmup.py:44] Skipping FlashInfer autotune because it is disabled.[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m (EngineCore_DP0 pid=940531) INFO 05-23 01:55:29 [core.py:278] init engine (profile, create kv cache, warmup model) took 2.88 seconds[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267) WARNING 05-23 01:55:28 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267) INFO 05-23 01:55:28 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267) WARNING 05-23 01:55:28 [vllm.py:735] Inductor compilation was disabled by user settings, optimizations settings that are only active during inductor compilation will be ignored.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=2552033, ip=10.128.43.88)[0m (EngineCore_DP0 pid=2552267) INFO 05-23 01:55:28 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=940292, ip=10.128.43.95)[0m WARNING 05-23 01:55:29 [model.py:1350] Default vLLM sampling parameters have been overridden by the model's `generation_config.json`: `{'temperature': 0.6, 'top_k': 20, 'top_p': 0.95}`. If this is not intended, please relaunch vLLM instance with `--generation-config vllm`.[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) INFO 05-23 01:55:30 [default_loader.py:293] Loading weights took 14.02 seconds[32m [repeated 11x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) INFO 05-23 01:55:30 [gpu_model_runner.py:4222] Model loading took 15.27 GiB memory and 15.199574 seconds[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank-0]: Successfully loaded model state dict
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976133) INFO 05-23 01:55:32 [gpu_worker.py:373] Available KV cache memory: 49.32 GiB[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976133) INFO 05-23 01:55:32 [kv_cache_utils.py:1307] GPU KV cache size: 359,104 tokens[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976133) INFO 05-23 01:55:32 [kv_cache_utils.py:1312] Maximum concurrency for 32,768 tokens per request: 10.96x[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976133) INFO 05-23 01:55:33 [kernel_warmup.py:44] Skipping FlashInfer autotune because it is disabled.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) INFO 05-23 01:55:33 [core.py:278] init engine (profile, create kv cache, warmup model) took 2.75 seconds[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) WARNING 05-23 01:55:33 [serial_utils.py:57] Allowing insecure serialization using pickle due to VLLM_ALLOW_INSECURE_SERIALIZATION=1[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) INFO 05-23 01:55:33 [vllm.py:690] Asynchronous scheduling is enabled.[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) WARNING 05-23 01:55:33 [vllm.py:735] Inductor compilation was disabled by user settings, optimizations settings that are only active during inductor compilation will be ignored.[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975888, ip=10.128.43.103)[0m (EngineCore_DP0 pid=976122) INFO 05-23 01:55:33 [vllm.py:846] Cudagraph is disabled under eager mode[32m [repeated 12x across cluster][0m
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=975890, ip=10.128.43.103)[0m WARNING 05-23 01:55:33 [model.py:1350] Default vLLM sampling parameters have been overridden by the model's `generation_config.json`: `{'temperature': 0.6, 'top_k': 20, 'top_p': 0.95}`. If this is not intended, please relaunch vLLM instance with `--generation-config vllm`.[32m [repeated 8x across cluster][0m
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank-0]: Successfully loaded optimizer state
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank-0]: Successfully loaded scheduler state
|
|||
|
|
[36m(FSDPPolicyWorkerBase pid=851205, ip=10.128.43.85)[0m [rank-0]: Checkpoint loaded successfully from /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints/global_step_18/policy
|
|||
|
|
[36m(AsyncVLLMInferenceEngine pid=937427, ip=10.128.43.84)[0m ERROR 05-23 01:55:45 [core_client.py:616] Engine core proc EngineCore_DP0 died unexpectedly, shutting down client.
|
|||
|
|
Error executing job with overrides: ['+terminal_bench_config=terminal_bench', 'trainer.strategy=fsdp2', 'trainer.algorithm.advantage_estimator=rloo_n', 'trainer.algorithm.use_kl_loss=false', 'trainer.algorithm.kl_loss_coef=0.0', 'trainer.algorithm.eps_clip_low=0.2', 'trainer.algorithm.eps_clip_high=0.05', 'trainer.algorithm.loss_reduction=token_mean', 'trainer.epochs=2', 'trainer.max_steps=80', 'trainer.update_epochs_per_batch=1', 'trainer.train_batch_size=64', 'trainer.policy_mini_batch_size=64', 'trainer.eval_batch_size=64', 'trainer.micro_forward_batch_size_per_gpu=4', 'trainer.micro_train_batch_size_per_gpu=1', 'trainer.max_prompt_length=999999', 'trainer.eval_interval=999999', 'trainer.eval_before_train=false', 'trainer.ckpt_interval=2', 'trainer.resume_mode=latest', 'trainer.hf_save_interval=5', '++trainer.hf_hub_repo_id=laion/a3-rl-DCAgent_code-contests-noblock', '++trainer.hf_hub_private=false', '++trainer.hf_hub_revision=main', '++trainer.enable_db_registration=true', 'trainer.project_name=OpenThoughts-Agent', 'trainer.log_level=INFO', 'trainer.tracker_commit_each_step=true', 'trainer.logger=console', 'trainer.run_name=a3-rl-DCAgent_code-contests-noblock', 'trainer.ckpt_path=/e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/checkpoints', 'trainer.export_path=/e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/exports', 'trainer.policy.optimizer_config.lr=8e-6', 'trainer.policy.optimizer_config.weight_decay=0.0', 'trainer.policy.optimizer_config.adam_betas=[0.9,0.999]', 'trainer.policy.optimizer_config.max_grad_norm=0.9', 'trainer.policy.fsdp_config.cpu_offload=false', 'trainer.policy.fsdp_config.reshard_after_forward=true', 'trainer.policy.fsdp_config.fsdp_size=4', 'trainer.policy.model.path=/e/data1/datasets/playground/ot-baf/hf_hub/models--laion--GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink/snapshots/0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', 'trainer.ref.fsdp_config.cpu_offload=false', 'trainer.ref.fsdp_config.reshard_after_forward=true', 'trainer.ref.fsdp_config.fsdp_size=4', 'trainer.placement.colocate_all=false', 'trainer.placement.policy_num_nodes=2', 'trainer.placement.ref_num_nodes=2', 'trainer.placement.policy_num_gpus_per_node=4', 'trainer.placement.ref_num_gpus_per_node=4', 'trainer.fully_async.max_staleness_steps=16', 'trainer.fully_async.num_parallel_generation_workers=450', 'generator.backend=vllm', 'generator.timeout_multiplier=1.0', 'generator.model_dtype=bfloat16', 'generator.inference_engine_tensor_parallel_size=1', 'generator.num_inference_engines=48', 'generator.n_samples_per_prompt=8', 'generator.eval_n_samples_per_prompt=8', 'generator.gpu_memory_utilization=0.75', 'generator.max_num_seqs=24', 'generator.max_num_batched_tokens=65536', 'generator.enable_prefix_caching=true', 'generator.enable_chunked_prefill=true', 'generator.run_engines_locally=true', 'generator.weight_sync_backend=nccl', 'generator.async_engine=true', 'generator.batched=false', 'generator.enable_http_endpoint=true', 'generator.enable_ray_prometheus_stats=false', 'generator.vllm_stats_interval=1', 'generator.append_eos_token_after_stop_str_in_multi_turn=true', 'generator.max_turns=999999', 'generator.sampling_params.max_generate_length=4096', 'generator.sampling_params.temperature=0.7', 'generator.sampling_params.top_p=0.95', 'generator.sampling_params.top_k=20', '++generator.engine_init_kwargs.max_model_len=32768', '++generator.engine_init_kwargs.custom_chat_template_chat_completion_path=chat_templates/qwen3_thinking_acc.jinja2', '++generator.engine_init_kwargs.served_model_name=0e3bff0c4e51f6b9ec0713b98b9eec36efb91cc6', 'data.train_data=["/e/scratch/jureap59/feuer1/tasks/code-contests-noblock"]', 'data.val_data=[]', '+terminal_bench_config.trials_dir=/e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock/trace_jobs', '+terminal_bench_config.harbor.name=terminus-2', '+terminal_bench_config.harbor.max_episodes=999999', '+te
|
|||
|
|
Traceback (most recent call last):
|
|||
|
|
File "<frozen runpy>", line 198, in _run_module_as_main
|
|||
|
|
File "<frozen runpy>", line 88, in _run_code
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/examples/terminal_bench/entrypoints/main_tbench.py", line 133, in <module>
|
|||
|
|
main()
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/main.py", line 94, in decorated_main
|
|||
|
|
_run_hydra(
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/_internal/utils.py", line 394, in _run_hydra
|
|||
|
|
_run_app(
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/_internal/utils.py", line 457, in _run_app
|
|||
|
|
run_and_report(
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/_internal/utils.py", line 223, in run_and_report
|
|||
|
|
raise ex
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/_internal/utils.py", line 220, in run_and_report
|
|||
|
|
return func()
|
|||
|
|
^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/_internal/utils.py", line 458, in <lambda>
|
|||
|
|
lambda: hydra.run(
|
|||
|
|
^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/_internal/hydra.py", line 132, in run
|
|||
|
|
_ = ret.return_value
|
|||
|
|
^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/core/utils.py", line 260, in return_value
|
|||
|
|
raise self._return_value
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/hydra/core/utils.py", line 186, in run_job
|
|||
|
|
ret.return_value = task_function(task_cfg)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/examples/terminal_bench/entrypoints/main_tbench.py", line 123, in main
|
|||
|
|
ray.get(skyrl_entrypoint.remote(cfg))
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/auto_init_hook.py", line 22, in auto_init_wrapper
|
|||
|
|
return fn(*args, **kwargs)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/client_mode_hook.py", line 104, in wrapper
|
|||
|
|
return func(*args, **kwargs)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/worker.py", line 2967, in get
|
|||
|
|
values, debugger_breakpoint = worker.get_objects(
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/ray/_private/worker.py", line 1015, in get_objects
|
|||
|
|
raise value.as_instanceof_cause()
|
|||
|
|
ray.exceptions.RayTaskError(ActorDiedError): [36mray::skyrl_entrypoint()[39m (pid=1461281, ip=10.128.43.81)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/examples/terminal_bench/entrypoints/main_tbench.py", line 96, in skyrl_entrypoint
|
|||
|
|
exp.run()
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/entrypoints/main_base.py", line 403, in run
|
|||
|
|
asyncio.run(trainer.train())
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/asyncio/runners.py", line 195, in run
|
|||
|
|
return runner.run(main)
|
|||
|
|
^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/asyncio/runners.py", line 118, in run
|
|||
|
|
return self._loop.run_until_complete(task)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/fully_async_trainer.py", line 434, in train
|
|||
|
|
await self._train_loop()
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/fully_async_trainer.py", line 488, in _train_loop
|
|||
|
|
self.init_weight_sync_state()
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/trainer.py", line 808, in init_weight_sync_state
|
|||
|
|
ray.get(
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
ray.exceptions.RayTaskError(ActorDiedError): [36mray::FSDPPolicyWorkerBase.init_weight_sync_state()[39m (pid=851205, ip=10.128.43.85, actor_id=8b87233f106bcf2285f12e1602000000, repr=<skyrl_train.workers.fsdp.fsdp_worker.FSDPPolicyWorkerBase object at 0x400dfe746e10>)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 449, in result
|
|||
|
|
return self.__get_result()
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
raise self._exception
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/workers/worker.py", line 299, in init_weight_sync_state
|
|||
|
|
results = await asyncio.gather(*tasks)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/inference_engine_client.py", line 655, in init_weight_update_communicator
|
|||
|
|
rank_offset_count += engine.tp_size() * engine.pp_size()
|
|||
|
|
^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/ray_wrapped_inference_engine.py", line 26, in tp_size
|
|||
|
|
return ray.get(self.inference_engine_actor.tp_size.remote())
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
ray.exceptions.ActorDiedError: The actor died because of an error raised in its creation task, [36mray::AsyncVLLMInferenceEngine.__init__()[39m (pid=876507, ip=10.128.43.96, actor_id=58494295fdaf9ec8226b54f502000000, repr=<skyrl_train.inference_engines.vllm.vllm_engine.AsyncVLLMInferenceEngine object at 0x400be192c230>)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 456, in result
|
|||
|
|
return self.__get_result()
|
|||
|
|
^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
|
|||
|
|
raise self._exception
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1028, in __init__
|
|||
|
|
super().__init__(*args, **kwargs)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 500, in __init__
|
|||
|
|
self.llm = self._create_engine(*args, **kwargs)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/SkyRL/skyrl-train/skyrl_train/inference_engines/vllm/vllm_engine.py", line 1081, in _create_engine
|
|||
|
|
engine = vllm.AsyncLLMEngine.from_engine_args(engine_args, stat_loggers=stat_loggers)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 251, in from_engine_args
|
|||
|
|
return cls(
|
|||
|
|
^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/async_llm.py", line 148, in __init__
|
|||
|
|
self.engine_core = EngineCoreClient.make_async_mp_client(
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 124, in make_async_mp_client
|
|||
|
|
return AsyncMPClient(*client_args)
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 835, in __init__
|
|||
|
|
super().__init__(
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/core_client.py", line 490, in __init__
|
|||
|
|
with launch_core_engines(vllm_config, executor_class, log_stats) as (
|
|||
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|||
|
|
File "/e/scratch/jureap59/feuer1/miniforge3/lib/python3.12/contextlib.py", line 144, in __exit__
|
|||
|
|
next(self.gen)
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 936, in launch_core_engines
|
|||
|
|
wait_for_engine_startup(
|
|||
|
|
File "/e/scratch/jureap59/feuer1/OpenThoughts-Agent/envs/rl/lib/python3.12/site-packages/vllm/v1/engine/utils.py", line 995, in wait_for_engine_startup
|
|||
|
|
raise RuntimeError(
|
|||
|
|
RuntimeError: Engine core initialization failed. See root cause above. Failed core proc(s): {}
|
|||
|
|
Stopping Ray cluster...
|
|||
|
|
Ray cluster stopped
|
|||
|
|
[RLJobRunner] Launching trace upload (training exit code: 1):
|
|||
|
|
repo_id: DCAgent/a3-rl-DCAgent_code-contests-noblock
|
|||
|
|
job_dir: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/a3-rl-DCAgent_code-contests-noblock
|
|||
|
|
episodes: last
|
|||
|
|
log: /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/logs/a3-rl-DCAgent_code-contests-noblock_trace_upload.log
|
|||
|
|
[RLJobRunner] Waiting for trace upload to complete...
|
|||
|
|
[RLJobRunner] Trace upload failed with exit code 1.
|
|||
|
|
Preserving Ray logs to /e/data1/datasets/playground/ot-baf/a3-rl-DCAgent_code-contests-noblock/ray_logs/
|
|||
|
|
Collecting Ray logs from worker jpbo-078-36...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-37...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-38...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-39...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-40...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-42...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-43...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-45...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-46...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-47...
|
|||
|
|
Collecting Ray logs from worker jpbo-078-48...
|
|||
|
|
Collecting Ray logs from worker jpbo-079-07...
|
|||
|
|
Collecting Ray logs from worker jpbo-079-12...
|
|||
|
|
Ray log preservation complete
|