g-h-chen commited on
Commit
033306b
1 Parent(s): 8812d34

upload config.json

Browse files
Files changed (1) hide show
  1. config.json +50 -0
config.json ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "FreedomIntelligence/ALLaVA-Phi2-2_7B",
3
+ "architectures": [
4
+ "LlavaPhiForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "auto_map": {
8
+ "AutoConfig": "configuration_phi.PhiConfig",
9
+ "AutoModelForCausalLM": "modeling_llava_phi.LlavaPhiForCausalLM"
10
+ },
11
+ "bos_token_id": null,
12
+ "embd_pdrop": 0.0,
13
+ "eos_token_id": null,
14
+ "flash_attn": true,
15
+ "flash_rotary": true,
16
+ "fused_dense": true,
17
+ "hidden_act": "gelu_new",
18
+ "hidden_size": 2560,
19
+ "image_aspect_ratio": "pad",
20
+ "initializer_range": 0.02,
21
+ "intermediate_size": 10240,
22
+ "layer_norm_eps": 1e-05,
23
+ "max_position_embeddings": 2048,
24
+ "mm_hidden_size": 1024,
25
+ "mm_projector_lr": null,
26
+ "mm_projector_type": "mlp2x_gelu",
27
+ "mm_use_im_patch_token": false,
28
+ "mm_use_im_start_end": false,
29
+ "mm_vision_select_feature": "patch",
30
+ "mm_vision_select_layer": -2,
31
+ "mm_vision_tower": "openai/clip-vit-large-patch14-336",
32
+ "model_type": "llava_phi",
33
+ "num_attention_heads": 32,
34
+ "num_hidden_layers": 32,
35
+ "num_key_value_heads": 32,
36
+ "partial_rotary_factor": 0.4,
37
+ "qk_layernorm": false,
38
+ "resid_pdrop": 0.1,
39
+ "rope_scaling": null,
40
+ "rope_theta": 10000.0,
41
+ "tie_word_embeddings": false,
42
+ "tokenizer_model_max_length": 2048,
43
+ "tokenizer_padding_side": "right",
44
+ "torch_dtype": "bfloat16",
45
+ "transformers_version": "4.37.0.dev0",
46
+ "tune_mm_mlp_adapter": false,
47
+ "use_cache": true,
48
+ "use_mm_proj": true,
49
+ "vocab_size": 51200
50
+ }