| { |
| "args": { |
| "config": null, |
| "dry_run": false, |
| "base": "Qwen/Qwen3.5-4B", |
| "backbone_adapter": "auto", |
| "branch_mode": "auto", |
| "lora_target_modules": "", |
| "base_load_path": "", |
| "epochs": 1, |
| "lr": 1e-05, |
| "head_lr": 0.0, |
| "weight_decay": 0.01, |
| "lora": 8, |
| "decision_mode": "pointer", |
| "rlcr": false, |
| "rlcr_group_size": 32, |
| "rlcr_sigma_start": 0.4, |
| "rlcr_sigma_end": 0.1, |
| "rlcr_policy_w": 1.0, |
| "rlcr_ce_w": 0.25, |
| "accum": 16, |
| "suite": "/l/users/huanxin.sheng/robo_train/release/robojudge-train-v0.2", |
| "device": "cuda", |
| "batch": 1, |
| "dtype": "bf16", |
| "weights_dtype": "bf16", |
| "checkpointing": 1, |
| "option_isolation": 0, |
| "special_embeddings": 0, |
| "multimodal": true, |
| "max_state": 4096, |
| "max_branch": 16384, |
| "max_packed": 16384, |
| "head_dim": 256, |
| "head_residual_dim": 256, |
| "lora_targets": "all", |
| "lora_dropout": 0.0, |
| "base_revision": "", |
| "p_none": 0.0, |
| "p_none_distract": 0.0, |
| "p_distract": 0.0, |
| "p_none_pair": 0.0, |
| "out": "/nfs-stor/huanxin.sheng/robo_train/runs/robojudge-4b-v02", |
| "data": "", |
| "replay": 0, |
| "init_from": "models/JevAny-Qwen3.5-4B-LoRA", |
| "seed": 0, |
| "max_steps": 0, |
| "eval_before_start": false, |
| "eval_every_steps": 1000, |
| "eval_records": 3000, |
| "eval_suite": "", |
| "eval_transfer_suite": "", |
| "checkpoint_every_steps": 1000, |
| "early_stop_patience": 0, |
| "early_stop_metric": "development_acc", |
| "early_stop_min_delta": 0.0, |
| "wandb_project": "", |
| "wandb_name": "", |
| "wandb_group": "", |
| "wandb_entity": "", |
| "wandb_mode": "disabled", |
| "resume": true, |
| "state_every_steps": 250, |
| "resume_allow_suite_change": true, |
| "time_budget_minutes": 160.0, |
| "deadline_minutes": 173.0 |
| }, |
| "suite_sha256": "94895633b5425e23b93d0f74144fdcbc48facfc27bad8059fccff96ac8b5ea5d", |
| "base_revision": "851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a", |
| "init_source": { |
| "init_from": "models/JevAny-Qwen3.5-4B-LoRA", |
| "resolved": "models/JevAny-Qwen3.5-4B-LoRA", |
| "adapter_sha256": "5acbaec8d554517250bf9c626b3f4b9e2ec31fc67957f3486aa72e2c91ee907f", |
| "head_sha256": "e5ea982efd06ecaf71b493045ebccdbfdca39f6fe128de164bf636d73152cd44", |
| "adapter_tensors": 496 |
| }, |
| "holdout": [], |
| "distributed": { |
| "world_size": 2, |
| "global_effective_batch": 32, |
| "per_rank_records": 92471, |
| "padding_records_per_epoch": 0 |
| } |
| } |
|
|
