wangrongsheng commited on
Commit
e2bbd4d
1 Parent(s): e5ccb72

Upload 2 files

Browse files
Files changed (2) hide show
  1. latest +1 -0
  2. model_config.json +73 -0
latest ADDED
@@ -0,0 +1 @@
 
 
1
+ 300
model_config.json ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_class": "FineTuneVisualGLMModel",
3
+ "tokenizer_type": "THUDM/chatglm-6b",
4
+ "num_layers": 28,
5
+ "hidden_size": 4096,
6
+ "num_attention_heads": 32,
7
+ "vocab_size": 130528,
8
+ "layernorm_order": "post",
9
+ "model_parallel_size": 1,
10
+ "max_sequence_length": 2048,
11
+ "pre_seq_len": 128,
12
+ "lora_rank": 10,
13
+ "use_ptuning": false,
14
+ "use_lora": true,
15
+ "image_length": 32,
16
+ "eva_args": {
17
+ "num_layers": 39,
18
+ "hidden_size": 1408,
19
+ "num_attention_heads": 16,
20
+ "vocab_size": 1,
21
+ "layernorm_order": "pre",
22
+ "model_parallel_size": 1,
23
+ "max_sequence_length": 257,
24
+ "inner_hidden_size": 6144,
25
+ "use_final_layernorm": false,
26
+ "layernorm_epsilon": 1e-06,
27
+ "image_size": [
28
+ 224,
29
+ 224
30
+ ],
31
+ "pre_len": 1,
32
+ "post_len": 0,
33
+ "in_channels": 3,
34
+ "num_classes": 0,
35
+ "patch_size": 14
36
+ },
37
+ "qformer_args": {
38
+ "num_layers": 12,
39
+ "hidden_size": 768,
40
+ "num_attention_heads": 12,
41
+ "vocab_size": 32,
42
+ "layernorm_order": "post",
43
+ "model_parallel_size": 1,
44
+ "max_sequence_length": 0,
45
+ "is_decoder": [
46
+ true,
47
+ false,
48
+ true,
49
+ false,
50
+ true,
51
+ false,
52
+ true,
53
+ false,
54
+ true,
55
+ false,
56
+ true,
57
+ false
58
+ ],
59
+ "cross_attn_hidden_size": 1408,
60
+ "layernorm_epsilon": 1e-12
61
+ },
62
+ "bos_token_id": 130004,
63
+ "mask_token_id": 130000,
64
+ "gmask_token_id": 130001,
65
+ "image_size": [
66
+ 224,
67
+ 224
68
+ ],
69
+ "pre_len": 1,
70
+ "post_len": 0,
71
+ "in_channels": 3,
72
+ "patch_size": 14
73
+ }