ThingsAI commited on
Commit
c5a9704
·
verified ·
1 Parent(s): fac5c1e

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +26 -10
config.json CHANGED
@@ -1,14 +1,30 @@
1
  {
 
 
2
  "vocab_size": 49152,
3
- "d_model": 384,
4
- "n_heads": 6,
5
- "n_kv_heads": 2,
6
- "n_layers": 24,
7
- "d_ff": 1024,
8
  "head_dim": 64,
9
- "max_seq_len": 2048,
 
 
 
10
  "rope_theta": 10000.0,
11
- "rms_eps": 1e-05,
12
- "qkv_bias": true,
13
- "dropout": 0.0
14
- }
 
 
 
 
 
 
 
 
 
 
 
 
1
  {
2
+ "architectures": ["LlamaForCausalLM"],
3
+ "model_type": "llama",
4
  "vocab_size": 49152,
5
+ "hidden_size": 384,
6
+ "intermediate_size": 1024,
7
+ "num_hidden_layers": 24,
8
+ "num_attention_heads": 6,
9
+ "num_key_value_heads": 2,
10
  "head_dim": 64,
11
+ "hidden_act": "silu",
12
+ "max_position_embeddings": 2048,
13
+ "initializer_range": 0.02,
14
+ "rms_norm_eps": 1e-05,
15
  "rope_theta": 10000.0,
16
+ "rope_scaling": null,
17
+ "rope_interleaved": false,
18
+ "attention_bias": true,
19
+ "attention_dropout": 0.0,
20
+ "mlp_bias": false,
21
+ "tie_word_embeddings": true,
22
+ "torch_dtype": "bfloat16",
23
+ "bos_token_id": 1,
24
+ "eos_token_id": 2,
25
+ "pad_token_id": 2,
26
+ "use_cache": true,
27
+ "pretraining_tp": 1,
28
+ "is_llama_config": true,
29
+ "transformers_version": "4.42.3"
30
+ }