Upload model_info.json with huggingface_hub
Browse files- model_info.json +29 -0
model_info.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"method": "relational_distillation_hard_neg",
|
| 3 |
+
"teacher": "intfloat/multilingual-e5-base",
|
| 4 |
+
"encoder_config": {
|
| 5 |
+
"vocab_size": 50000,
|
| 6 |
+
"hidden": 384,
|
| 7 |
+
"layers": 12,
|
| 8 |
+
"heads": 12,
|
| 9 |
+
"ffn": 1536,
|
| 10 |
+
"dropout": 0.1,
|
| 11 |
+
"max_position": 512,
|
| 12 |
+
"pooling": "mean",
|
| 13 |
+
"activation": "gelu"
|
| 14 |
+
},
|
| 15 |
+
"has_proj": true,
|
| 16 |
+
"param_count": {
|
| 17 |
+
"total": 40986624,
|
| 18 |
+
"trainable": 40986624,
|
| 19 |
+
"non_trainable": 0,
|
| 20 |
+
"fp32_mb": 156.4,
|
| 21 |
+
"bf16_fp16_mb": 78.2
|
| 22 |
+
},
|
| 23 |
+
"final_eval": {
|
| 24 |
+
"val_rel_mse": 0.0371,
|
| 25 |
+
"val_acc1": 0.8494
|
| 26 |
+
},
|
| 27 |
+
"hard_negatives": true,
|
| 28 |
+
"pad_id": 0
|
| 29 |
+
}
|