Xidong commited on
Commit
8f730fc
1 Parent(s): 6027891

Upload ./config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +49 -0
config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/sds_wangby/models/Jamba-v0.1",
3
+ "architectures": [
4
+ "JambaForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "attn_layer_offset": 4,
8
+ "attn_layer_period": 8,
9
+ "auto_map": {
10
+ "AutoConfig": "configuration_jamba.JambaConfig",
11
+ "AutoModel": "modeling_jamba.JambaModel",
12
+ "AutoModelForCausalLM": "modeling_jamba.JambaForCausalLM",
13
+ "AutoModelForSequenceClassification": "model.JambaForSequenceClassification"
14
+ },
15
+ "bos_token_id": 1,
16
+ "calc_logits_for_entire_prompt": false,
17
+ "eos_token_id": 2,
18
+ "expert_layer_offset": 1,
19
+ "expert_layer_period": 2,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 14336,
24
+ "mamba_conv_bias": true,
25
+ "mamba_d_conv": 4,
26
+ "mamba_d_state": 16,
27
+ "mamba_dt_rank": 256,
28
+ "mamba_expand": 2,
29
+ "mamba_inner_layernorms": true,
30
+ "mamba_proj_bias": false,
31
+ "model_type": "jamba",
32
+ "n_ctx": 262144,
33
+ "num_attention_heads": 32,
34
+ "num_experts": 16,
35
+ "num_experts_per_tok": 2,
36
+ "num_hidden_layers": 32,
37
+ "num_key_value_heads": 8,
38
+ "output_router_logits": false,
39
+ "pad_token_id": 0,
40
+ "rms_norm_eps": 1e-06,
41
+ "router_aux_loss_coef": 0.001,
42
+ "sliding_window": null,
43
+ "tie_word_embeddings": false,
44
+ "torch_dtype": "bfloat16",
45
+ "transformers_version": "4.39.3",
46
+ "use_cache": true,
47
+ "use_mamba_kernels": true,
48
+ "vocab_size": 65536
49
+ }