TheBloke commited on
Commit
2fc923b
1 Parent(s): d0b6929

Update for Transformers AWQ support

Browse files
Files changed (1) hide show
  1. config.json +38 -31
config.json CHANGED
@@ -1,33 +1,40 @@
1
  {
2
- "_name_or_path": "OpenAssistant/codellama-13b-oasst-sft-v10",
3
- "architectures": [
4
- "LlamaForCausalLM"
5
- ],
6
- "auto_map": {
7
- "AutoConfig": "configuration_llama.LlamaConfig",
8
- "AutoModel": "modeling_llama.LlamaModel",
9
- "AutoModelForCausalLM": "modeling_llama.LlamaForCausalLM",
10
- "AutoModelForSequenceClassification": "modeling_llama.LlamaForSequenceClassification"
11
- },
12
- "bos_token_id": 32021,
13
- "eos_token_id": 32022,
14
- "hidden_act": "silu",
15
- "hidden_size": 5120,
16
- "initializer_range": 0.02,
17
- "intermediate_size": 13824,
18
- "max_position_embeddings": 16384,
19
- "model_type": "llama",
20
- "num_attention_heads": 40,
21
- "num_hidden_layers": 40,
22
- "num_key_value_heads": 40,
23
- "pad_token_id": 0,
24
- "pretraining_tp": 1,
25
- "rms_norm_eps": 1e-05,
26
- "rope_scaling": null,
27
- "rope_theta": 1000000,
28
- "tie_word_embeddings": false,
29
- "torch_dtype": "bfloat16",
30
- "transformers_version": "4.31.0",
31
- "use_cache": true,
32
- "vocab_size": 32032
 
 
 
 
 
 
 
33
  }
 
1
  {
2
+ "_name_or_path": "OpenAssistant/codellama-13b-oasst-sft-v10",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "auto_map": {
7
+ "AutoConfig": "configuration_llama.LlamaConfig",
8
+ "AutoModel": "modeling_llama.LlamaModel",
9
+ "AutoModelForCausalLM": "modeling_llama.LlamaForCausalLM",
10
+ "AutoModelForSequenceClassification": "modeling_llama.LlamaForSequenceClassification"
11
+ },
12
+ "bos_token_id": 32021,
13
+ "eos_token_id": 32022,
14
+ "hidden_act": "silu",
15
+ "hidden_size": 5120,
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 13824,
18
+ "max_position_embeddings": 16384,
19
+ "model_type": "llama",
20
+ "num_attention_heads": 40,
21
+ "num_hidden_layers": 40,
22
+ "num_key_value_heads": 40,
23
+ "pad_token_id": 0,
24
+ "pretraining_tp": 1,
25
+ "rms_norm_eps": 1e-05,
26
+ "rope_scaling": null,
27
+ "rope_theta": 1000000,
28
+ "tie_word_embeddings": false,
29
+ "torch_dtype": "bfloat16",
30
+ "transformers_version": "4.31.0",
31
+ "use_cache": true,
32
+ "vocab_size": 32032,
33
+ "quantization_config": {
34
+ "quant_method": "awq",
35
+ "zero_point": true,
36
+ "group_size": 128,
37
+ "bits": 4,
38
+ "version": "gemm"
39
+ }
40
  }