periodic: runs/exp_t1_fusion_sft_unified_ntv3_lora_n200k_20260428_h100/_arch_meta.json
Browse files
runs/exp_t1_fusion_sft_unified_ntv3_lora_n200k_20260428_h100/_arch_meta.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"task": "T1",
|
| 3 |
+
"out_name": "exp_t1_fusion_sft_unified_ntv3_lora_n200k_20260428_h100",
|
| 4 |
+
"arch_mode": "unified",
|
| 5 |
+
"plan": "B",
|
| 6 |
+
"train_size": "200k (35k strat7c + 165k random from full enriched v2 enh-scan)",
|
| 7 |
+
"speed_knobs": {
|
| 8 |
+
"gradient_checkpointing": false,
|
| 9 |
+
"eval_steps": 2000,
|
| 10 |
+
"batch_size": 8,
|
| 11 |
+
"grad_accum": 1,
|
| 12 |
+
"apply_liger": true
|
| 13 |
+
},
|
| 14 |
+
"qwen_lora_r": 8,
|
| 15 |
+
"ntv3_lora_r": 16,
|
| 16 |
+
"max_length": 5120,
|
| 17 |
+
"epochs": 1,
|
| 18 |
+
"lr": "2e-5",
|
| 19 |
+
"started_utc": "2026-04-28T12:29:07Z",
|
| 20 |
+
"label": "T1 fusion-SFT — Plan B headline (unified_ntv3_lora, 200k mixed strat7c+random + frozen reasoning)"
|
| 21 |
+
}
|