SeongryongJung/Qwen3-8B-Chemistry-RLRT-TR
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: SeongryongJung/Qwen3-8B-Chemistry-RLRT-TR
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-10-03 20:59:23 +08:00
commit 404189931e
21 changed files with 152411 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
artifacts/config.yaml Normal file
View File

@@ -0,0 +1,18 @@
{
"experiment_name": "qwen3gen-chemistry-RLRT_TR-Qwen-Qwen3-8B-mbs8-tr0.1-decay0-train32-rollout8-lr1e-6-vllm0.8-epsw0.5",
"base_model": "Qwen/Qwen3-8B",
"dataset": "chemistry",
"method": "RLRT-TR",
"checkpoint_dir": "/mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLRT_TR-Qwen-Qwen3-8B-mbs8-tr0.1-decay0-train32-rollout8-lr1e-6-vllm0.8-epsw0.5",
"best_checkpoint": "global_step_100",
"last_checkpoint": "global_step_100",
"train_batch_size": 32,
"train_max_samples": 3200,
"teacher_regularization": "trust-region",
"teacher_update_rate": 0.1,
"rollout_n": 8,
"validation_n": 16,
"learning_rate": "1e-6",
"vllm_gpu_memory_utilization": 0.8,
"command_args": ""
}

10
artifacts/wandb-summary.json Normal file
View File

@@ -0,0 +1,10 @@
{
"source": "parsed local training log",
"experiment_name": "qwen3gen-chemistry-RLRT_TR-Qwen-Qwen3-8B-mbs8-tr0.1-decay0-train32-rollout8-lr1e-6-vllm0.8-epsw0.5",
"best_val_mean16": 0.0,
"best_step": 100,
"final_val_mean16": 0.0,
"final_step": 100,
"uploaded_root_checkpoint": "global_step_100",
"uploaded_last_checkpoint": "global_step_100"
}