Update config.json
Browse files- config.json +25 -69
config.json
CHANGED
@@ -5,9 +5,13 @@
|
|
5 |
"attention_dropout": 0.0,
|
6 |
"bos_token_id": 151643,
|
7 |
"eos_token_id": 151645,
|
|
|
|
|
|
|
|
|
|
|
8 |
"hidden_act": "silu",
|
9 |
"hidden_size": 3584,
|
10 |
-
"image_token_id": 151655,
|
11 |
"initializer_range": 0.02,
|
12 |
"intermediate_size": 18944,
|
13 |
"max_position_embeddings": 128000,
|
@@ -39,89 +43,41 @@
|
|
39 |
]
|
40 |
},
|
41 |
"rms_norm_eps": 1e-06,
|
42 |
-
"rope_scaling": {
|
43 |
-
"mrope_section": [
|
44 |
-
16,
|
45 |
-
24,
|
46 |
-
24
|
47 |
-
],
|
48 |
-
"rope_type": "default",
|
49 |
-
"type": "default"
|
50 |
-
},
|
51 |
"rope_theta": 1000000.0,
|
52 |
"sliding_window": 32768,
|
53 |
-
"text_config": {
|
54 |
-
"architectures": [
|
55 |
-
"Qwen2_5_VLForConditionalGeneration"
|
56 |
-
],
|
57 |
-
"attention_dropout": 0.0,
|
58 |
-
"bos_token_id": 151643,
|
59 |
-
"eos_token_id": 151645,
|
60 |
-
"hidden_act": "silu",
|
61 |
-
"hidden_size": 3584,
|
62 |
-
"image_token_id": null,
|
63 |
-
"initializer_range": 0.02,
|
64 |
-
"intermediate_size": 18944,
|
65 |
-
"max_position_embeddings": 128000,
|
66 |
-
"max_window_layers": 28,
|
67 |
-
"model_type": "qwen2_5_vl_text",
|
68 |
-
"num_attention_heads": 28,
|
69 |
-
"num_hidden_layers": 28,
|
70 |
-
"num_key_value_heads": 4,
|
71 |
-
"rms_norm_eps": 1e-06,
|
72 |
-
"rope_scaling": {
|
73 |
-
"mrope_section": [
|
74 |
-
16,
|
75 |
-
24,
|
76 |
-
24
|
77 |
-
],
|
78 |
-
"rope_type": "default",
|
79 |
-
"type": "default"
|
80 |
-
},
|
81 |
-
"rope_theta": 1000000.0,
|
82 |
-
"sliding_window": 32768,
|
83 |
-
"torch_dtype": "float16",
|
84 |
-
"use_cache": true,
|
85 |
-
"use_sliding_window": false,
|
86 |
-
"video_token_id": null,
|
87 |
-
"vision_end_token_id": 151653,
|
88 |
-
"vision_start_token_id": 151652,
|
89 |
-
"vision_token_id": 151654,
|
90 |
-
"vocab_size": 152064
|
91 |
-
},
|
92 |
"tie_word_embeddings": false,
|
93 |
-
"torch_dtype": "
|
94 |
-
"transformers_version": "4.
|
95 |
"use_cache": true,
|
96 |
"use_sliding_window": false,
|
97 |
-
"video_token_id": 151656,
|
98 |
"vision_config": {
|
99 |
"depth": 32,
|
100 |
-
"fullatt_block_indexes": [
|
101 |
-
7,
|
102 |
-
15,
|
103 |
-
23,
|
104 |
-
31
|
105 |
-
],
|
106 |
"hidden_act": "silu",
|
107 |
"hidden_size": 1280,
|
108 |
-
"in_channels": 3,
|
109 |
-
"in_chans": 3,
|
110 |
-
"initializer_range": 0.02,
|
111 |
"intermediate_size": 3420,
|
112 |
-
"model_type": "qwen2_5_vl",
|
113 |
"num_heads": 16,
|
|
|
114 |
"out_hidden_size": 3584,
|
115 |
"patch_size": 14,
|
116 |
"spatial_merge_size": 2,
|
117 |
"spatial_patch_size": 14,
|
118 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
119 |
"tokens_per_second": 2,
|
120 |
-
"
|
121 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
122 |
},
|
123 |
-
"vision_end_token_id": 151653,
|
124 |
-
"vision_start_token_id": 151652,
|
125 |
-
"vision_token_id": 151654,
|
126 |
"vocab_size": 152064
|
127 |
-
}
|
|
|
5 |
"attention_dropout": 0.0,
|
6 |
"bos_token_id": 151643,
|
7 |
"eos_token_id": 151645,
|
8 |
+
"vision_start_token_id": 151652,
|
9 |
+
"vision_end_token_id": 151653,
|
10 |
+
"vision_token_id": 151654,
|
11 |
+
"image_token_id": 151655,
|
12 |
+
"video_token_id": 151656,
|
13 |
"hidden_act": "silu",
|
14 |
"hidden_size": 3584,
|
|
|
15 |
"initializer_range": 0.02,
|
16 |
"intermediate_size": 18944,
|
17 |
"max_position_embeddings": 128000,
|
|
|
43 |
]
|
44 |
},
|
45 |
"rms_norm_eps": 1e-06,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
"rope_theta": 1000000.0,
|
47 |
"sliding_window": 32768,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
"tie_word_embeddings": false,
|
49 |
+
"torch_dtype": "bfloat16",
|
50 |
+
"transformers_version": "4.41.2",
|
51 |
"use_cache": true,
|
52 |
"use_sliding_window": false,
|
|
|
53 |
"vision_config": {
|
54 |
"depth": 32,
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
"hidden_act": "silu",
|
56 |
"hidden_size": 1280,
|
|
|
|
|
|
|
57 |
"intermediate_size": 3420,
|
|
|
58 |
"num_heads": 16,
|
59 |
+
"in_chans": 3,
|
60 |
"out_hidden_size": 3584,
|
61 |
"patch_size": 14,
|
62 |
"spatial_merge_size": 2,
|
63 |
"spatial_patch_size": 14,
|
64 |
+
"window_size": 112,
|
65 |
+
"fullatt_block_indexes": [
|
66 |
+
7,
|
67 |
+
15,
|
68 |
+
23,
|
69 |
+
31
|
70 |
+
],
|
71 |
"tokens_per_second": 2,
|
72 |
+
"temporal_patch_size": 2
|
73 |
+
},
|
74 |
+
"rope_scaling": {
|
75 |
+
"type": "mrope",
|
76 |
+
"mrope_section": [
|
77 |
+
16,
|
78 |
+
24,
|
79 |
+
24
|
80 |
+
]
|
81 |
},
|
|
|
|
|
|
|
82 |
"vocab_size": 152064
|
83 |
+
}
|