Sri-Vigneshwar-DJ commited on
Commit
10b2dc2
1 Parent(s): 6428a3c

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +62 -0
config.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "meta-llama/Llama-3.3-70B-Instruct",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 128000,
9
+ "eos_token_id": [
10
+ 128001,
11
+ 128008,
12
+ 128009
13
+ ],
14
+ "head_dim": 128,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 8192,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 28672,
19
+ "max_position_embeddings": 131072,
20
+ "mlp_bias": false,
21
+ "model_type": "llama",
22
+ "num_attention_heads": 64,
23
+ "num_hidden_layers": 80,
24
+ "num_key_value_heads": 8,
25
+ "pad_token_id": 128004,
26
+ "pretraining_tp": 1,
27
+ "quantization_config": {
28
+ "_load_in_4bit": true,
29
+ "_load_in_8bit": false,
30
+ "bnb_4bit_compute_dtype": "bfloat16",
31
+ "bnb_4bit_quant_storage": "uint8",
32
+ "bnb_4bit_quant_type": "nf4",
33
+ "bnb_4bit_use_double_quant": true,
34
+ "llm_int8_enable_fp32_cpu_offload": false,
35
+ "llm_int8_has_fp16_weight": false,
36
+ "llm_int8_skip_modules": [
37
+ "lm_head",
38
+ "multi_modal_projector",
39
+ "merger",
40
+ "modality_projection"
41
+ ],
42
+ "llm_int8_threshold": 6.0,
43
+ "load_in_4bit": true,
44
+ "load_in_8bit": false,
45
+ "quant_method": "bitsandbytes"
46
+ },
47
+ "rms_norm_eps": 1e-05,
48
+ "rope_scaling": {
49
+ "factor": 8.0,
50
+ "high_freq_factor": 4.0,
51
+ "low_freq_factor": 1.0,
52
+ "original_max_position_embeddings": 8192,
53
+ "rope_type": "llama3"
54
+ },
55
+ "rope_theta": 500000.0,
56
+ "tie_word_embeddings": false,
57
+ "torch_dtype": "bfloat16",
58
+ "transformers_version": "4.47.0",
59
+ "unsloth_fixed": true,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }