maghrane commited on
Commit
8d09643
1 Parent(s): 0598775

Upload config.json

Browse files
Files changed (1) hide show
  1. config.json +73 -0
config.json ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openbmb/MiniCPM-Llama3-V-2_5-int4",
3
+ "architectures": [
4
+ "MiniCPMV"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "auto_map": {
9
+ "AutoConfig": "configuration_minicpm.MiniCPMVConfig",
10
+ "AutoModel": "modeling_minicpmv.MiniCPMV",
11
+ "AutoModelForCausalLM": "modeling_minicpmv.MiniCPMV"
12
+ },
13
+ "batch_vision_input": true,
14
+ "bos_token_id": 128000,
15
+ "drop_vision_last_layer": false,
16
+ "eos_token_id": 128001,
17
+ "hidden_act": "silu",
18
+ "hidden_size": 4096,
19
+ "image_size": 448,
20
+ "initializer_range": 0.02,
21
+ "intermediate_size": 14336,
22
+ "max_position_embeddings": 8192,
23
+ "mm_use_im_start_end": true,
24
+ "model_type": "minicpmv",
25
+ "num_attention_heads": 32,
26
+ "num_hidden_layers": 32,
27
+ "num_key_value_heads": 8,
28
+ "patch_size": 14,
29
+ "pretraining_tp": 1,
30
+ "quantization_config": {
31
+ "_load_in_4bit": true,
32
+ "_load_in_8bit": false,
33
+ "bnb_4bit_compute_dtype": "float16",
34
+ "bnb_4bit_quant_storage": "uint8",
35
+ "bnb_4bit_quant_type": "nf4",
36
+ "bnb_4bit_use_double_quant": true,
37
+ "llm_int8_enable_fp32_cpu_offload": false,
38
+ "llm_int8_has_fp16_weight": false,
39
+ "llm_int8_skip_modules": [
40
+ "out_proj",
41
+ "kv_proj",
42
+ "lm_head"
43
+ ],
44
+ "llm_int8_threshold": 6.0,
45
+ "load_in_4bit": true,
46
+ "load_in_8bit": false,
47
+ "quant_method": "bitsandbytes"
48
+ },
49
+ "query_num": 96,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": null,
52
+ "rope_theta": 500000.0,
53
+ "slice_config": {
54
+ "max_slice_nums": 9,
55
+ "patch_size": 14,
56
+ "model_type": "minicpmv"
57
+ },
58
+ "slice_mode": true,
59
+ "tie_word_embeddings": false,
60
+ "torch_dtype": "float16",
61
+ "transformers_version": "4.40.0",
62
+ "use_cache": false,
63
+ "vision_config": {
64
+ "hidden_size": 1152,
65
+ "image_size": 980,
66
+ "intermediate_size": 4304,
67
+ "model_type": "idefics2",
68
+ "num_attention_heads": 16,
69
+ "num_hidden_layers": 27,
70
+ "patch_size": 14
71
+ },
72
+ "vocab_size": 128256
73
+ }