Upload config.json with huggingface_hub
Browse files- config.json +37 -0
config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LongcatFlashOmniForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"auto_map": {
|
| 8 |
+
"AutoConfig": "configuration_longcat_flash.LongcatFlashConfig",
|
| 9 |
+
"AutoModel": "modeling_longcat_flash.LongcatFlashModel",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_longcat_flash.LongcatFlashForCausalLM"
|
| 11 |
+
},
|
| 12 |
+
"vocab_size": 131072,
|
| 13 |
+
"hidden_size": 6144,
|
| 14 |
+
"ffn_hidden_size": 12288,
|
| 15 |
+
"expert_ffn_hidden_size": 2048,
|
| 16 |
+
"num_layers": 28,
|
| 17 |
+
"num_attention_heads": 64,
|
| 18 |
+
"kv_lora_rank": 512,
|
| 19 |
+
"q_lora_rank": 1536,
|
| 20 |
+
"qk_rope_head_dim": 64,
|
| 21 |
+
"v_head_dim": 128,
|
| 22 |
+
"qk_nope_head_dim": 128,
|
| 23 |
+
"mla_scale_q_lora": true,
|
| 24 |
+
"mla_scale_kv_lora": true,
|
| 25 |
+
"routed_scaling_factor": 6.0,
|
| 26 |
+
"n_routed_experts": 512,
|
| 27 |
+
"max_position_embeddings": 131072,
|
| 28 |
+
"rms_norm_eps": 1e-5,
|
| 29 |
+
"use_cache": true,
|
| 30 |
+
"bos_token_id": 1,
|
| 31 |
+
"eos_token_id": 2,
|
| 32 |
+
"rope_theta": 10000000.0,
|
| 33 |
+
"attention_method": "MLA",
|
| 34 |
+
"zero_expert_num": 256,
|
| 35 |
+
"zero_expert_type": "identity",
|
| 36 |
+
"moe_topk": 12
|
| 37 |
+
}
|