Commit
·
a417bc8
1
Parent(s):
e131816
修改Qwen config
Browse files- .gitignore +1 -0
- Qwen-14B-Chat/bert4torch_config.json +2 -4
- Qwen-14B/bert4torch_config.json +1 -3
- Qwen-1_8B-Chat/bert4torch_config.json +4 -12
- Qwen-1_8B/bert4torch_config.json +3 -11
- Qwen-7B-Chat/bert4torch_config.json +3 -11
- Qwen-7B/bert4torch_config.json +2 -10
- Qwen1.5-0.5B-Chat/bert4torch_config.json +1 -1
- Qwen1.5-1.8B-Chat/bert4torch_config.json +1 -1
- Qwen1.5-14B-Chat/bert4torch_config.json +1 -1
- Qwen1.5-7B-Chat/bert4torch_config.json +1 -1
- Qwen2-0.5B-Instruct/bert4torch_config.json +1 -1
- Qwen2-1.5B-Instruct/bert4torch_config.json +1 -1
- Qwen2-7B-Instruct/bert4torch_config.json +1 -1
.gitignore
CHANGED
@@ -1 +1,2 @@
|
|
1 |
config.json
|
|
|
|
1 |
config.json
|
2 |
+
generation_config.json
|
Qwen-14B-Chat/bert4torch_config.json
CHANGED
@@ -7,12 +7,10 @@
|
|
7 |
"hidden_size": 5120,
|
8 |
"num_attention_heads": 40,
|
9 |
"num_hidden_layers": 40,
|
10 |
-
"
|
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 |
"rope_scaling": {"type": "dynamic_qwen"},
|
@@ -22,6 +20,6 @@
|
|
22 |
"rope_rank": "updown",
|
23 |
"max_position_embeddings": 8192,
|
24 |
"generation_config": {"tokenizer_config": {"allowed_special": ["<|im_end|>", "<|im_start|>", "<|endoftext|>"],
|
25 |
-
"skip_special_tokens": true}, "eos_token_id": [151644, 151645],
|
26 |
"max_length": 8192}
|
27 |
}
|
|
|
7 |
"hidden_size": 5120,
|
8 |
"num_attention_heads": 40,
|
9 |
"num_hidden_layers": 40,
|
10 |
+
"rope_theta": 10000,
|
|
|
11 |
"scale_attn_weights": true,
|
12 |
"seq_length": 2048,
|
13 |
"tie_word_embeddings": false,
|
|
|
14 |
"flash_attention": "flash_attn_2",
|
15 |
"vocab_size": 152064,
|
16 |
"rope_scaling": {"type": "dynamic_qwen"},
|
|
|
20 |
"rope_rank": "updown",
|
21 |
"max_position_embeddings": 8192,
|
22 |
"generation_config": {"tokenizer_config": {"allowed_special": ["<|im_end|>", "<|im_start|>", "<|endoftext|>"],
|
23 |
+
"skip_special_tokens": true}, "eos_token_id": [151643, 151644, 151645],
|
24 |
"max_length": 8192}
|
25 |
}
|
Qwen-14B/bert4torch_config.json
CHANGED
@@ -7,12 +7,10 @@
|
|
7 |
"hidden_size": 5120,
|
8 |
"num_attention_heads": 40,
|
9 |
"num_hidden_layers": 40,
|
10 |
-
"
|
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 |
"rope_scaling": {"type": "dynamic_qwen"},
|
|
|
7 |
"hidden_size": 5120,
|
8 |
"num_attention_heads": 40,
|
9 |
"num_hidden_layers": 40,
|
10 |
+
"rope_theta": 10000,
|
|
|
11 |
"scale_attn_weights": true,
|
12 |
"seq_length": 2048,
|
13 |
"tie_word_embeddings": false,
|
|
|
14 |
"flash_attention": "flash_attn_2",
|
15 |
"vocab_size": 152064,
|
16 |
"rope_scaling": {"type": "dynamic_qwen"},
|
Qwen-1_8B-Chat/bert4torch_config.json
CHANGED
@@ -1,26 +1,18 @@
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
-
"
|
5 |
-
"bos_token_id": 151643,
|
6 |
-
"embd_pdrop": 0.1,
|
7 |
"eos_token_id": 151643,
|
8 |
"intermediate_size": 5504,
|
9 |
"initializer_range": 0.02,
|
10 |
-
"kv_channels": 128,
|
11 |
"layer_norm_eps": 1e-06,
|
12 |
"hidden_size": 2048,
|
13 |
"num_attention_heads": 16,
|
14 |
"num_hidden_layers": 24,
|
15 |
-
"
|
16 |
-
"resid_pdrop": 0.1,
|
17 |
-
"rotary_emb_base": 10000,
|
18 |
-
"rotary_pct": 1.0,
|
19 |
"scale_attn_weights": true,
|
20 |
-
"seq_length": 8192,
|
21 |
"tie_word_embeddings": false,
|
22 |
-
"
|
23 |
-
"use_flash_attn": true,
|
24 |
"vocab_size": 151936,
|
25 |
"rope_scaling": {"type": "dynamic_qwen"},
|
26 |
"use_logn_attn": true,
|
@@ -29,6 +21,6 @@
|
|
29 |
"rope_rank": "updown",
|
30 |
"max_position": 8192,
|
31 |
"generation_config": {"tokenizer_config": {"allowed_special": ["<|im_end|>", "<|im_start|>", "<|endoftext|>"],
|
32 |
-
"skip_special_tokens": true}, "eos_token_id": [151644, 151645],
|
33 |
"max_length": 8192}
|
34 |
}
|
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
+
"pad_token_id": 151643,
|
|
|
|
|
5 |
"eos_token_id": 151643,
|
6 |
"intermediate_size": 5504,
|
7 |
"initializer_range": 0.02,
|
|
|
8 |
"layer_norm_eps": 1e-06,
|
9 |
"hidden_size": 2048,
|
10 |
"num_attention_heads": 16,
|
11 |
"num_hidden_layers": 24,
|
12 |
+
"rope_theta": 10000,
|
|
|
|
|
|
|
13 |
"scale_attn_weights": true,
|
|
|
14 |
"tie_word_embeddings": false,
|
15 |
+
"flash_attention": true,
|
|
|
16 |
"vocab_size": 151936,
|
17 |
"rope_scaling": {"type": "dynamic_qwen"},
|
18 |
"use_logn_attn": true,
|
|
|
21 |
"rope_rank": "updown",
|
22 |
"max_position": 8192,
|
23 |
"generation_config": {"tokenizer_config": {"allowed_special": ["<|im_end|>", "<|im_start|>", "<|endoftext|>"],
|
24 |
+
"skip_special_tokens": true}, "eos_token_id": [151643, 151644, 151645],
|
25 |
"max_length": 8192}
|
26 |
}
|
Qwen-1_8B/bert4torch_config.json
CHANGED
@@ -1,26 +1,18 @@
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
-
"
|
5 |
-
"bos_token_id": 151643,
|
6 |
-
"embd_pdrop": 0.1,
|
7 |
"eos_token_id": 151643,
|
8 |
"intermediate_size": 5504,
|
9 |
"initializer_range": 0.02,
|
10 |
-
"kv_channels": 128,
|
11 |
"layer_norm_eps": 1e-06,
|
12 |
"hidden_size": 2048,
|
13 |
"num_attention_heads": 16,
|
14 |
"num_hidden_layers": 24,
|
15 |
-
"
|
16 |
-
"resid_pdrop": 0.1,
|
17 |
-
"rotary_emb_base": 10000,
|
18 |
-
"rotary_pct": 1.0,
|
19 |
"scale_attn_weights": true,
|
20 |
-
"seq_length": 8192,
|
21 |
"tie_word_embeddings": false,
|
22 |
-
"
|
23 |
-
"use_flash_attn": true,
|
24 |
"vocab_size": 151936,
|
25 |
"rope_scaling": {"type": "dynamic_qwen"},
|
26 |
"use_logn_attn": true,
|
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
+
"pad_token_id": 151643,
|
|
|
|
|
5 |
"eos_token_id": 151643,
|
6 |
"intermediate_size": 5504,
|
7 |
"initializer_range": 0.02,
|
|
|
8 |
"layer_norm_eps": 1e-06,
|
9 |
"hidden_size": 2048,
|
10 |
"num_attention_heads": 16,
|
11 |
"num_hidden_layers": 24,
|
12 |
+
"rope_theta": 10000,
|
|
|
|
|
|
|
13 |
"scale_attn_weights": true,
|
|
|
14 |
"tie_word_embeddings": false,
|
15 |
+
"flash_attention": true,
|
|
|
16 |
"vocab_size": 151936,
|
17 |
"rope_scaling": {"type": "dynamic_qwen"},
|
18 |
"use_logn_attn": true,
|
Qwen-7B-Chat/bert4torch_config.json
CHANGED
@@ -1,25 +1,17 @@
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
-
"
|
5 |
-
"bos_token_id": 151643,
|
6 |
-
"embd_pdrop": 0.1,
|
7 |
"eos_token_id": 151643,
|
8 |
"intermediate_size": 11008,
|
9 |
"initializer_range": 0.02,
|
10 |
-
"kv_channels": 128,
|
11 |
"layer_norm_eps": 1e-05,
|
12 |
"hidden_size": 4096,
|
13 |
"num_attention_heads": 32,
|
14 |
"num_hidden_layers": 32,
|
15 |
-
"
|
16 |
-
"resid_pdrop": 0.1,
|
17 |
-
"rotary_emb_base": 10000,
|
18 |
-
"rotary_pct": 1.0,
|
19 |
"scale_attn_weights": true,
|
20 |
-
"seq_length": 8192,
|
21 |
"tie_word_embeddings": false,
|
22 |
-
"use_cache": true,
|
23 |
"flash_attention": "flash_attn_2",
|
24 |
"vocab_size": 151936,
|
25 |
"rope_scaling": {"type": "dynamic_qwen"},
|
@@ -29,6 +21,6 @@
|
|
29 |
"rope_rank": "updown",
|
30 |
"max_position": 8192,
|
31 |
"generation_config": {"tokenizer_config": {"allowed_special": ["<|im_end|>", "<|im_start|>", "<|endoftext|>"],
|
32 |
-
"skip_special_tokens": true}, "eos_token_id": [151644, 151645],
|
33 |
"max_length": 8192}
|
34 |
}
|
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
+
"pad_token_id": 151643,
|
|
|
|
|
5 |
"eos_token_id": 151643,
|
6 |
"intermediate_size": 11008,
|
7 |
"initializer_range": 0.02,
|
|
|
8 |
"layer_norm_eps": 1e-05,
|
9 |
"hidden_size": 4096,
|
10 |
"num_attention_heads": 32,
|
11 |
"num_hidden_layers": 32,
|
12 |
+
"rope_theta": 10000,
|
|
|
|
|
|
|
13 |
"scale_attn_weights": true,
|
|
|
14 |
"tie_word_embeddings": false,
|
|
|
15 |
"flash_attention": "flash_attn_2",
|
16 |
"vocab_size": 151936,
|
17 |
"rope_scaling": {"type": "dynamic_qwen"},
|
|
|
21 |
"rope_rank": "updown",
|
22 |
"max_position": 8192,
|
23 |
"generation_config": {"tokenizer_config": {"allowed_special": ["<|im_end|>", "<|im_start|>", "<|endoftext|>"],
|
24 |
+
"skip_special_tokens": true}, "eos_token_id": [151643, 151644, 151645],
|
25 |
"max_length": 8192}
|
26 |
}
|
Qwen-7B/bert4torch_config.json
CHANGED
@@ -1,25 +1,17 @@
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
-
"
|
5 |
-
"bos_token_id": 151643,
|
6 |
-
"embd_pdrop": 0.1,
|
7 |
"eos_token_id": 151643,
|
8 |
"intermediate_size": 11008,
|
9 |
"initializer_range": 0.02,
|
10 |
-
"kv_channels": 128,
|
11 |
"layer_norm_eps": 1e-05,
|
12 |
"hidden_size": 4096,
|
13 |
"num_attention_heads": 32,
|
14 |
"num_hidden_layers": 32,
|
15 |
-
"
|
16 |
-
"resid_pdrop": 0.1,
|
17 |
-
"rotary_emb_base": 10000,
|
18 |
-
"rotary_pct": 1.0,
|
19 |
"scale_attn_weights": true,
|
20 |
-
"seq_length": 8192,
|
21 |
"tie_word_embeddings": false,
|
22 |
-
"use_cache": true,
|
23 |
"flash_attention": "flash_attn_2",
|
24 |
"vocab_size": 151936,
|
25 |
"rope_scaling": {"type": "dynamic_qwen"},
|
|
|
1 |
{
|
2 |
"model": "qwen",
|
3 |
"hidden_act": "silu",
|
4 |
+
"pad_token_id": 151643,
|
|
|
|
|
5 |
"eos_token_id": 151643,
|
6 |
"intermediate_size": 11008,
|
7 |
"initializer_range": 0.02,
|
|
|
8 |
"layer_norm_eps": 1e-05,
|
9 |
"hidden_size": 4096,
|
10 |
"num_attention_heads": 32,
|
11 |
"num_hidden_layers": 32,
|
12 |
+
"rope_theta": 10000,
|
|
|
|
|
|
|
13 |
"scale_attn_weights": true,
|
|
|
14 |
"tie_word_embeddings": false,
|
|
|
15 |
"flash_attention": "flash_attn_2",
|
16 |
"vocab_size": 151936,
|
17 |
"rope_scaling": {"type": "dynamic_qwen"},
|
Qwen1.5-0.5B-Chat/bert4torch_config.json
CHANGED
@@ -24,6 +24,6 @@
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 21,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
-
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [
|
28 |
"max_length": 32768}
|
29 |
}
|
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 21,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
+
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
|
28 |
"max_length": 32768}
|
29 |
}
|
Qwen1.5-1.8B-Chat/bert4torch_config.json
CHANGED
@@ -24,6 +24,6 @@
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 21,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
-
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [
|
28 |
"max_length": 32768}
|
29 |
}
|
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 21,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
+
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
|
28 |
"max_length": 32768}
|
29 |
}
|
Qwen1.5-14B-Chat/bert4torch_config.json
CHANGED
@@ -24,6 +24,6 @@
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 35,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
-
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [
|
28 |
"max_length": 32768}
|
29 |
}
|
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 35,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
+
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
|
28 |
"max_length": 32768}
|
29 |
}
|
Qwen1.5-7B-Chat/bert4torch_config.json
CHANGED
@@ -24,6 +24,6 @@
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 28,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
-
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [
|
28 |
"max_length": 32768}
|
29 |
}
|
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 28,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
+
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
|
28 |
"max_length": 32768}
|
29 |
}
|
Qwen2-0.5B-Instruct/bert4torch_config.json
CHANGED
@@ -24,6 +24,6 @@
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 24,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
-
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [
|
28 |
"max_length": 32768}
|
29 |
}
|
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 24,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
+
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
|
28 |
"max_length": 32768}
|
29 |
}
|
Qwen2-1.5B-Instruct/bert4torch_config.json
CHANGED
@@ -24,6 +24,6 @@
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 28,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
-
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [
|
28 |
"max_length": 32768}
|
29 |
}
|
|
|
24 |
"sliding_window": 32768,
|
25 |
"max_window_layers": 28,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
+
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
|
28 |
"max_length": 32768}
|
29 |
}
|
Qwen2-7B-Instruct/bert4torch_config.json
CHANGED
@@ -24,6 +24,6 @@
|
|
24 |
"sliding_window": 131072,
|
25 |
"max_window_layers": 28,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
-
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [
|
28 |
"max_length": 32768}
|
29 |
}
|
|
|
24 |
"sliding_window": 131072,
|
25 |
"max_window_layers": 28,
|
26 |
"convert_lm_logits_dtype": "float32",
|
27 |
+
"generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
|
28 |
"max_length": 32768}
|
29 |
}
|