| { | |
| "model_name": "Qwen/Qwen3-1.7B", | |
| "hf_repo_id": "farpluto/doc-to-lora-niah", | |
| "lora_r": 8, | |
| "lora_alpha": 8.0, | |
| "target_module": "down_proj", | |
| "latent_dim": 512, | |
| "n_percv_blocks": 8, | |
| "lr": 4e-05, | |
| "weight_decay": 0.01, | |
| "grad_accum": 8, | |
| "max_steps": 8000, | |
| "warmup_steps": 200, | |
| "max_grad_norm": 1.0, | |
| "l1_coef": 0.05, | |
| "eval_every": 500, | |
| "save_every": 2000, | |
| "ctx_min_len": 32, | |
| "ctx_max_len": 256, | |
| "n_train": 80000, | |
| "n_eval": 300, | |
| "seed": 42, | |
| "out_dir": "d2l_ckpts" | |
| } |