RichardErkhov commited on
Commit
81e8d37
1 Parent(s): fb3174e

uploaded model

Browse files
Files changed (1) hide show
  1. config.json +49 -0
config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "fuyu-8b",
3
+ "architectures": [
4
+ "FuyuForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 71013,
9
+ "hidden_act": "relu2",
10
+ "hidden_dropout": 0.0,
11
+ "hidden_size": 4096,
12
+ "image_size": 300,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 16384,
15
+ "layer_norm_eps": 1e-05,
16
+ "max_position_embeddings": 16384,
17
+ "model_type": "fuyu",
18
+ "num_attention_heads": 64,
19
+ "num_channels": 3,
20
+ "num_hidden_layers": 36,
21
+ "partial_rotary_factor": 0.5,
22
+ "patch_size": 30,
23
+ "qk_layernorm": true,
24
+ "quantization_config": {
25
+ "_load_in_4bit": true,
26
+ "_load_in_8bit": false,
27
+ "bnb_4bit_compute_dtype": "float32",
28
+ "bnb_4bit_quant_storage": "uint8",
29
+ "bnb_4bit_quant_type": "fp4",
30
+ "bnb_4bit_use_double_quant": false,
31
+ "llm_int8_enable_fp32_cpu_offload": false,
32
+ "llm_int8_has_fp16_weight": false,
33
+ "llm_int8_skip_modules": null,
34
+ "llm_int8_threshold": 6.0,
35
+ "load_in_4bit": true,
36
+ "load_in_8bit": false,
37
+ "quant_method": "bitsandbytes"
38
+ },
39
+ "rope_scaling": null,
40
+ "rope_theta": 25000.0,
41
+ "text_config": {
42
+ "model_type": "persimmon"
43
+ },
44
+ "tie_word_embeddings": false,
45
+ "torch_dtype": "float16",
46
+ "transformers_version": "4.40.2",
47
+ "use_cache": true,
48
+ "vocab_size": 262144
49
+ }