Tongjilibo
commited on
Commit
•
d73ebe7
1
Parent(s):
08601ed
增加qwen14b
Browse files
Qwen-14B-Chat/bert4torch_config.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"model": "qwen",
|
3 |
+
"hidden_act": "silu",
|
4 |
+
"intermediate_size": 13696,
|
5 |
+
"initializer_range": 0.02,
|
6 |
+
"layer_norm_eps": 1e-06,
|
7 |
+
"hidden_size": 5120,
|
8 |
+
"num_attention_heads": 40,
|
9 |
+
"num_hidden_layers": 40,
|
10 |
+
"rotary_emb_base": 10000,
|
11 |
+
"rotary_pct": 1.0,
|
12 |
+
"scale_attn_weights": true,
|
13 |
+
"seq_length": 2048,
|
14 |
+
"tie_word_embeddings": false,
|
15 |
+
"use_cache": true,
|
16 |
+
"flash_attention": "flash_attn_2",
|
17 |
+
"vocab_size": 152064,
|
18 |
+
"use_dynamic_ntk": true,
|
19 |
+
"use_logn_attn": true,
|
20 |
+
"segment_vocab_size": 0,
|
21 |
+
"skip_init": true,
|
22 |
+
"rope_rank": "updown",
|
23 |
+
"max_position_embeddings": 8192
|
24 |
+
}
|
Qwen-14B/bert4torch_config.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"model": "qwen",
|
3 |
+
"hidden_act": "silu",
|
4 |
+
"intermediate_size": 13696,
|
5 |
+
"initializer_range": 0.02,
|
6 |
+
"layer_norm_eps": 1e-06,
|
7 |
+
"hidden_size": 5120,
|
8 |
+
"num_attention_heads": 40,
|
9 |
+
"num_hidden_layers": 40,
|
10 |
+
"rotary_emb_base": 10000,
|
11 |
+
"rotary_pct": 1.0,
|
12 |
+
"scale_attn_weights": true,
|
13 |
+
"seq_length": 2048,
|
14 |
+
"tie_word_embeddings": false,
|
15 |
+
"use_cache": true,
|
16 |
+
"flash_attention": "flash_attn_2",
|
17 |
+
"vocab_size": 152064,
|
18 |
+
"use_dynamic_ntk": true,
|
19 |
+
"use_logn_attn": true,
|
20 |
+
"segment_vocab_size": 0,
|
21 |
+
"skip_init": true,
|
22 |
+
"rope_rank": "updown",
|
23 |
+
"max_position_embeddings": 8192
|
24 |
+
}
|