Upload training_manifest.json with huggingface_hub
Browse files- training_manifest.json +46 -0
training_manifest.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"schema": "sov33.training-manifest/v1",
|
| 3 |
+
"lora_adapter": {
|
| 4 |
+
"base_model": "Qwen/Qwen2.5-3B-Instruct",
|
| 5 |
+
"training_examples": 3967,
|
| 6 |
+
"epochs": 3,
|
| 7 |
+
"lora_rank": 32,
|
| 8 |
+
"lora_alpha": 64,
|
| 9 |
+
"trainable_params": 59867136,
|
| 10 |
+
"total_params": 3145805824,
|
| 11 |
+
"training_time_seconds": 1433,
|
| 12 |
+
"final_loss": 0.0
|
| 13 |
+
},
|
| 14 |
+
"data_sources": {
|
| 15 |
+
"sov33_original": 561,
|
| 16 |
+
"dolly_15k": 2000,
|
| 17 |
+
"alpaca_gpt4": 982,
|
| 18 |
+
"hf_alpaca": 500,
|
| 19 |
+
"codealpaca": 485
|
| 20 |
+
},
|
| 21 |
+
"benchmark_results": {
|
| 22 |
+
"sov_ultimate_modelfile": {
|
| 23 |
+
"overall": 0.95
|
| 24 |
+
},
|
| 25 |
+
"sov6v2": {
|
| 26 |
+
"overall": 0.867
|
| 27 |
+
},
|
| 28 |
+
"qwen2.5_3b_base": {
|
| 29 |
+
"overall": 0.62
|
| 30 |
+
}
|
| 31 |
+
},
|
| 32 |
+
"govbench_v8": {
|
| 33 |
+
"qwen2.5_3b": {
|
| 34 |
+
"accuracy": 1.0,
|
| 35 |
+
"overblock": 0.0
|
| 36 |
+
},
|
| 37 |
+
"sov6v2": {
|
| 38 |
+
"accuracy": 1.0,
|
| 39 |
+
"overblock": 0.0
|
| 40 |
+
},
|
| 41 |
+
"sov5v2": {
|
| 42 |
+
"accuracy": 0.982,
|
| 43 |
+
"overblock": 0.1
|
| 44 |
+
}
|
| 45 |
+
}
|
| 46 |
+
}
|