Osaurus-AI commited on
Commit
8cb8b5d
·
verified ·
1 Parent(s): 70b2d41

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +74 -0
config.json ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BailingMoeV2_5ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "auto_map": {
7
+ "AutoConfig": "configuration_bailing_moe_v2_5.BailingMoeV2_5Config",
8
+ "AutoModel": "modeling_bailing_moe_v2_5.BailingMoeV2_5Model",
9
+ "AutoModelForCausalLM": "modeling_bailing_moe_v2_5.BailingMoeV2_5ForCausalLM"
10
+ },
11
+ "embedding_dropout": 0.0,
12
+ "eos_token_id": 156895,
13
+ "first_k_dense_replace": 1,
14
+ "group_norm_size": 4,
15
+ "head_dim": 128,
16
+ "hidden_act": "silu",
17
+ "hidden_size": 4096,
18
+ "initializer_range": 0.02,
19
+ "intermediate_size": 9216,
20
+ "kv_lora_rank": 512,
21
+ "layer_group_size": 8,
22
+ "linear_silu": false,
23
+ "max_position_embeddings": 131072,
24
+ "max_window_layers": 20,
25
+ "moe_intermediate_size": 1024,
26
+ "moe_router_enable_expert_bias": true,
27
+ "moe_shared_expert_intermediate_size": 1024,
28
+ "mtp_loss_scaling_factor": 0,
29
+ "n_group": 8,
30
+ "num_attention_heads": 32,
31
+ "num_experts": 256,
32
+ "num_experts_per_tok": 8,
33
+ "num_hidden_layers": 32,
34
+ "num_key_value_heads": 32,
35
+ "num_kv_heads_for_linear_attn": 32,
36
+ "num_nextn_predict_layers": 1,
37
+ "num_shared_experts": 1,
38
+ "output_dropout": 0.0,
39
+ "output_router_logits": false,
40
+ "pad_token_id": 156892,
41
+ "partial_rotary_factor": 0.5,
42
+ "q_lora_rank": 1536,
43
+ "qk_head_dim": 192,
44
+ "qk_nope_head_dim": 128,
45
+ "qk_rope_head_dim": 64,
46
+ "rms_norm_eps": 1e-06,
47
+ "rope_interleave": true,
48
+ "rope_scaling": null,
49
+ "rope_theta": 6000000,
50
+ "rotary_dim": 64,
51
+ "routed_scaling_factor": 2.5,
52
+ "router_dtype": "fp32",
53
+ "score_function": "sigmoid",
54
+ "scoring_func": "sigmoid",
55
+ "seq_aux": true,
56
+ "tie_word_embeddings": false,
57
+ "topk_group": 4,
58
+ "topk_method": "noaux_tc",
59
+ "transformers_version": "4.56.2",
60
+ "use_bias": false,
61
+ "use_cache": true,
62
+ "use_qk_norm": true,
63
+ "use_qkv_bias": false,
64
+ "v_head_dim": 128,
65
+ "vocab_size": 157184,
66
+ "model_type": "bailing_hybrid",
67
+ "torch_dtype": "bfloat16",
68
+ "quantization": {
69
+ "bits": 4,
70
+ "group_size": 32,
71
+ "mode": "affine"
72
+ },
73
+ "weight_format": "mxfp4"
74
+ }