danielhanchen commited on
Commit
1714feb
1 Parent(s): 70b6a51

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +4 -13
config.json CHANGED
@@ -1,28 +1,21 @@
1
  {
2
  "_name_or_path": "unsloth/Phi-3-medium-4k-instruct",
3
  "architectures": [
4
- "Phi3ForCausalLM"
5
  ],
6
- "attention_bias": false,
7
  "attention_dropout": 0.0,
8
- "auto_map": {
9
- "AutoConfig": "microsoft/Phi-3-medium-4k-instruct--configuration_phi3.Phi3Config",
10
- "AutoModelForCausalLM": "microsoft/Phi-3-medium-4k-instruct--modeling_phi3.Phi3ForCausalLM"
11
- },
12
  "bos_token_id": 1,
13
- "embd_pdrop": 0.0,
14
  "eos_token_id": 32000,
15
  "hidden_act": "silu",
16
  "hidden_size": 5120,
17
  "initializer_range": 0.02,
18
  "intermediate_size": 17920,
19
  "max_position_embeddings": 4096,
20
- "model_type": "phi3",
21
  "num_attention_heads": 40,
22
  "num_hidden_layers": 40,
23
  "num_key_value_heads": 10,
24
- "original_max_position_embeddings": 4096,
25
- "pad_token_id": 32000,
26
  "quantization_config": {
27
  "_load_in_4bit": true,
28
  "_load_in_8bit": false,
@@ -38,11 +31,9 @@
38
  "load_in_8bit": false,
39
  "quant_method": "bitsandbytes"
40
  },
41
- "resid_pdrop": 0.0,
42
  "rms_norm_eps": 1e-05,
43
- "rope_scaling": null,
44
  "rope_theta": 10000.0,
45
- "sliding_window": 2047,
46
  "tie_word_embeddings": false,
47
  "torch_dtype": "bfloat16",
48
  "transformers_version": "4.44.0",
 
1
  {
2
  "_name_or_path": "unsloth/Phi-3-medium-4k-instruct",
3
  "architectures": [
4
+ "MistralForCausalLM"
5
  ],
 
6
  "attention_dropout": 0.0,
 
 
 
 
7
  "bos_token_id": 1,
 
8
  "eos_token_id": 32000,
9
  "hidden_act": "silu",
10
  "hidden_size": 5120,
11
  "initializer_range": 0.02,
12
  "intermediate_size": 17920,
13
  "max_position_embeddings": 4096,
14
+ "model_type": "mistral",
15
  "num_attention_heads": 40,
16
  "num_hidden_layers": 40,
17
  "num_key_value_heads": 10,
18
+ "pad_token_id": 32009,
 
19
  "quantization_config": {
20
  "_load_in_4bit": true,
21
  "_load_in_8bit": false,
 
31
  "load_in_8bit": false,
32
  "quant_method": "bitsandbytes"
33
  },
 
34
  "rms_norm_eps": 1e-05,
 
35
  "rope_theta": 10000.0,
36
+ "sliding_window": 2048,
37
  "tie_word_embeddings": false,
38
  "torch_dtype": "bfloat16",
39
  "transformers_version": "4.44.0",