Add config
Browse files- config.json +24 -0
config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_name": "Qwen/Qwen3-1.7B",
|
| 3 |
+
"hf_repo_id": "farpluto/doc-to-lora-niah",
|
| 4 |
+
"lora_r": 8,
|
| 5 |
+
"lora_alpha": 8.0,
|
| 6 |
+
"target_module": "down_proj",
|
| 7 |
+
"latent_dim": 512,
|
| 8 |
+
"n_percv_blocks": 8,
|
| 9 |
+
"lr": 4e-05,
|
| 10 |
+
"weight_decay": 0.01,
|
| 11 |
+
"grad_accum": 8,
|
| 12 |
+
"max_steps": 8000,
|
| 13 |
+
"warmup_steps": 200,
|
| 14 |
+
"max_grad_norm": 1.0,
|
| 15 |
+
"l1_coef": 0.05,
|
| 16 |
+
"eval_every": 500,
|
| 17 |
+
"save_every": 2000,
|
| 18 |
+
"ctx_min_len": 32,
|
| 19 |
+
"ctx_max_len": 256,
|
| 20 |
+
"n_train": 80000,
|
| 21 |
+
"n_eval": 300,
|
| 22 |
+
"seed": 42,
|
| 23 |
+
"out_dir": "d2l_ckpts"
|
| 24 |
+
}
|