Add model config
Browse files- config.json +37 -0
config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architecture": "HRM-Text (classification)",
|
| 3 |
+
"reference": "sapientinc/HRM-Text, arXiv:2506.21734",
|
| 4 |
+
"hidden_size": 768,
|
| 5 |
+
"num_heads": 12,
|
| 6 |
+
"head_dim": 64,
|
| 7 |
+
"n_layers_H": 3,
|
| 8 |
+
"n_layers_L": 3,
|
| 9 |
+
"intermediate_size": 2048,
|
| 10 |
+
"H_cycles": 2,
|
| 11 |
+
"L_cycles": 3,
|
| 12 |
+
"max_seq_len": 2048,
|
| 13 |
+
"rope_base": 10000.0,
|
| 14 |
+
"rope_scaling": 1.0,
|
| 15 |
+
"bp_min_steps": 2,
|
| 16 |
+
"bp_max_steps": 5,
|
| 17 |
+
"vocab_size": 256,
|
| 18 |
+
"param_count": 46206722,
|
| 19 |
+
"id2label": {
|
| 20 |
+
"0": "safe",
|
| 21 |
+
"1": "injection"
|
| 22 |
+
},
|
| 23 |
+
"label2id": {
|
| 24 |
+
"safe": 0,
|
| 25 |
+
"injection": 1
|
| 26 |
+
},
|
| 27 |
+
"training": {
|
| 28 |
+
"learning_rate": 0.00022,
|
| 29 |
+
"epochs": 1,
|
| 30 |
+
"batch_size": 32,
|
| 31 |
+
"weight_decay": 0.1,
|
| 32 |
+
"scheduler": "constant_with_warmup",
|
| 33 |
+
"warmup_steps": 2000,
|
| 34 |
+
"adam_beta2": 0.95,
|
| 35 |
+
"precision": "bf16"
|
| 36 |
+
}
|
| 37 |
+
}
|