Upload run_summary.json with huggingface_hub
Browse files- run_summary.json +34 -0
run_summary.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"completed_at": "2026-03-31T16:18:39Z",
|
| 3 |
+
"model_id": "Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2",
|
| 4 |
+
"device": "cuda",
|
| 5 |
+
"engine": "unsloth",
|
| 6 |
+
"dtype": "torch.bfloat16",
|
| 7 |
+
"max_steps": 2928,
|
| 8 |
+
"requested_max_steps": 100,
|
| 9 |
+
"completed_steps": 2928,
|
| 10 |
+
"final_eval_loss": 0.5050073671564786,
|
| 11 |
+
"train_dataset_size": 11712,
|
| 12 |
+
"eval_dataset_size": 1465,
|
| 13 |
+
"sequence_length": 4096,
|
| 14 |
+
"learning_rate": 5e-05,
|
| 15 |
+
"gradient_accumulation": 4,
|
| 16 |
+
"train_epochs": 1.0,
|
| 17 |
+
"update_steps_per_epoch": 2928,
|
| 18 |
+
"effective_epochs": 1.0,
|
| 19 |
+
"batch_size": 1,
|
| 20 |
+
"load_in_4bit": true,
|
| 21 |
+
"lora_target_modules": [
|
| 22 |
+
"down_proj",
|
| 23 |
+
"gate_proj",
|
| 24 |
+
"k_proj",
|
| 25 |
+
"o_proj",
|
| 26 |
+
"q_proj",
|
| 27 |
+
"up_proj",
|
| 28 |
+
"v_proj"
|
| 29 |
+
],
|
| 30 |
+
"parameter_summary": {
|
| 31 |
+
"total_parameters": 15135298800,
|
| 32 |
+
"trainable_parameters": 159383552
|
| 33 |
+
}
|
| 34 |
+
}
|