Upload config.json with huggingface_hub
Browse files- config.json +29 -0
config.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"DeeplmModel"
|
| 4 |
+
],
|
| 5 |
+
"model_type": "deeplm",
|
| 6 |
+
"vocab_size": 32000,
|
| 7 |
+
"hidden_size": 512,
|
| 8 |
+
"intermediate_size": 2048,
|
| 9 |
+
"num_hidden_layers": 10,
|
| 10 |
+
"num_attention_heads": 8,
|
| 11 |
+
"num_key_value_heads": 1,
|
| 12 |
+
"max_position_embeddings": 4096,
|
| 13 |
+
"rms_norm_eps": 1e-06,
|
| 14 |
+
"rope_theta": 50000.0,
|
| 15 |
+
"rope_dim": 64,
|
| 16 |
+
"tie_word_embeddings": true,
|
| 17 |
+
"num_routed_experts": 4,
|
| 18 |
+
"num_shared_experts": 1,
|
| 19 |
+
"expert_topk": 2,
|
| 20 |
+
"q_lora_rank": 192,
|
| 21 |
+
"kv_lora_rank": 64,
|
| 22 |
+
"qk_rope_head_dim": 64,
|
| 23 |
+
"qk_nope_head_dim": 64,
|
| 24 |
+
"v_head_dim": 128,
|
| 25 |
+
"mtp_depth": 2,
|
| 26 |
+
"mtp_num_layers": 2,
|
| 27 |
+
"bitnet_quantized": true,
|
| 28 |
+
"bitnet_scale": "absmean"
|
| 29 |
+
}
|