Files changed (1) hide show
  1. config.json +12 -15
config.json CHANGED
@@ -1,30 +1,27 @@
1
  {
2
  "_name_or_path": "/home/ubuntu/smaug",
3
- "architectures": [
4
- "LlamaForCausalLM"
5
  ],
6
- "attention_bias": true,
7
  "attention_dropout": 0.0,
8
- "bos_token_id": 1,
9
- "eos_token_id": 151643,
10
  "hidden_act": "silu",
11
  "hidden_size": 8192,
12
  "initializer_range": 0.02,
13
  "intermediate_size": 24576,
14
  "max_position_embeddings": 32768,
15
- "model_type": "llama",
 
16
  "num_attention_heads": 64,
17
- "num_hidden_layers": 85,
18
  "num_key_value_heads": 64,
19
- "pad_token_id": 151643,
20
- "pretraining_tp": 1,
21
- "rms_norm_eps": 1e-06,
22
- "rope_scaling": null,
23
- "rope_theta": 1000000,
24
- "seq_length": 32768,
25
  "tie_word_embeddings": false,
26
  "torch_dtype": "float16",
27
- "transformers_version": "4.38.1",
28
- "use_cache": true,
 
29
  "vocab_size": 152064
30
  }
 
1
  {
2
  "_name_or_path": "/home/ubuntu/smaug",
3
+ "architectures": [
4
+ "Qwen2ForCausalLM"
5
  ],
 
6
  "attention_dropout": 0.0,
7
+ "eos_token_id": 151645,
 
8
  "hidden_act": "silu",
9
  "hidden_size": 8192,
10
  "initializer_range": 0.02,
11
  "intermediate_size": 24576,
12
  "max_position_embeddings": 32768,
13
+ "max_window_layers": 28,
14
+ "model_type": "qwen2",
15
  "num_attention_heads": 64,
16
+ "num_hidden_layers": 80,
17
  "num_key_value_heads": 64,
18
+ "rms_norm_eps": 1e-05,
19
+ "rope_theta": 1000000.0,
20
+ "sliding_window": 32768,
 
 
 
21
  "tie_word_embeddings": false,
22
  "torch_dtype": "float16",
23
+ "transformers_version": "4.39.0.dev0",
24
+ "use_cache": false,
25
+ "use_sliding_window": false,
26
  "vocab_size": 152064
27
  }