YaTharThShaRma999 commited on
Commit
dab51d8
·
verified ·
1 Parent(s): 11086bb

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +7 -11
config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "architectures": [
3
- "LlamaModel"
4
  ],
5
  "attention_bias": false,
6
  "attention_dropout": 0.0,
@@ -11,26 +11,22 @@
11
  "hidden_size": 1024,
12
  "initializer_range": 0.02,
13
  "intermediate_size": 8192,
14
- "max_position_embeddings": 32,
15
  "mlp_bias": false,
16
  "model_type": "llama",
17
  "num_attention_heads": 8,
18
  "num_hidden_layers": 4,
19
  "num_key_value_heads": 2,
20
- "pretraining_tp": 1,
21
  "rms_norm_eps": 1e-05,
22
  "rope_scaling": {
23
  "factor": 32.0,
24
- "high_freq_factor": 4.0,
25
- "low_freq_factor": 1.0,
26
- "original_max_position_embeddings": 8192,
27
- "rope_type": "llama3",
28
- "type": "llama3"
29
  },
30
  "rope_theta": 500000,
31
  "tie_word_embeddings": false,
32
- "torch_dtype": "float16",
33
- "transformers_version": "4.51.3",
34
  "use_cache": true,
35
- "vocab_size": 128256
36
  }
 
1
  {
2
  "architectures": [
3
+ "LlamaForCausalLM"
4
  ],
5
  "attention_bias": false,
6
  "attention_dropout": 0.0,
 
11
  "hidden_size": 1024,
12
  "initializer_range": 0.02,
13
  "intermediate_size": 8192,
14
+ "max_position_embeddings": 33,
15
  "mlp_bias": false,
16
  "model_type": "llama",
17
  "num_attention_heads": 8,
18
  "num_hidden_layers": 4,
19
  "num_key_value_heads": 2,
 
20
  "rms_norm_eps": 1e-05,
21
  "rope_scaling": {
22
  "factor": 32.0,
23
+ "high_freq_factor": 0.0078125,
24
+ "low_freq_factor": 0.001953125,
25
+ "original_max_position_embeddings": 16,
26
+ "rope_type": "llama3"
 
27
  },
28
  "rope_theta": 500000,
29
  "tie_word_embeddings": false,
 
 
30
  "use_cache": true,
31
+ "vocab_size": 2051
32
  }