Initial commit
Browse files- config.json +18 -3
- generation_config.json +1 -1
- model-00001-of-00004.safetensors +2 -2
- model-00002-of-00004.safetensors +2 -2
- model-00003-of-00004.safetensors +2 -2
- model-00004-of-00004.safetensors +2 -2
- model.safetensors.index.json +0 -0
- preprocessor_config.json +4 -0
config.json
CHANGED
@@ -1,5 +1,4 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "../checkpoints/openvlthinker_grpo_iter3_hard/global_step_30/actor/huggingface",
|
3 |
"architectures": [
|
4 |
"Qwen2_5_VLForConditionalGeneration"
|
5 |
],
|
@@ -31,17 +30,33 @@
|
|
31 |
"sliding_window": 32768,
|
32 |
"tie_word_embeddings": false,
|
33 |
"torch_dtype": "float32",
|
34 |
-
"transformers_version": "4.
|
35 |
"use_cache": false,
|
36 |
"use_sliding_window": false,
|
37 |
"video_token_id": 151656,
|
38 |
"vision_config": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
"hidden_size": 1280,
|
|
|
40 |
"in_chans": 3,
|
|
|
41 |
"model_type": "qwen2_5_vl",
|
|
|
|
|
|
|
|
|
42 |
"spatial_patch_size": 14,
|
|
|
43 |
"tokens_per_second": 2,
|
44 |
-
"torch_dtype": "float32"
|
|
|
45 |
},
|
46 |
"vision_end_token_id": 151653,
|
47 |
"vision_start_token_id": 151652,
|
|
|
1 |
{
|
|
|
2 |
"architectures": [
|
3 |
"Qwen2_5_VLForConditionalGeneration"
|
4 |
],
|
|
|
30 |
"sliding_window": 32768,
|
31 |
"tie_word_embeddings": false,
|
32 |
"torch_dtype": "float32",
|
33 |
+
"transformers_version": "4.51.0",
|
34 |
"use_cache": false,
|
35 |
"use_sliding_window": false,
|
36 |
"video_token_id": 151656,
|
37 |
"vision_config": {
|
38 |
+
"depth": 32,
|
39 |
+
"fullatt_block_indexes": [
|
40 |
+
7,
|
41 |
+
15,
|
42 |
+
23,
|
43 |
+
31
|
44 |
+
],
|
45 |
+
"hidden_act": "silu",
|
46 |
"hidden_size": 1280,
|
47 |
+
"in_channels": 3,
|
48 |
"in_chans": 3,
|
49 |
+
"intermediate_size": 3420,
|
50 |
"model_type": "qwen2_5_vl",
|
51 |
+
"num_heads": 16,
|
52 |
+
"out_hidden_size": 3584,
|
53 |
+
"patch_size": 14,
|
54 |
+
"spatial_merge_size": 2,
|
55 |
"spatial_patch_size": 14,
|
56 |
+
"temporal_patch_size": 2,
|
57 |
"tokens_per_second": 2,
|
58 |
+
"torch_dtype": "float32",
|
59 |
+
"window_size": 112
|
60 |
},
|
61 |
"vision_end_token_id": 151653,
|
62 |
"vision_start_token_id": 151652,
|
generation_config.json
CHANGED
@@ -2,6 +2,6 @@
|
|
2 |
"_from_model_config": true,
|
3 |
"eos_token_id": 151645,
|
4 |
"pad_token_id": 151643,
|
5 |
-
"transformers_version": "4.
|
6 |
"use_cache": false
|
7 |
}
|
|
|
2 |
"_from_model_config": true,
|
3 |
"eos_token_id": 151645,
|
4 |
"pad_token_id": 151643,
|
5 |
+
"transformers_version": "4.51.0",
|
6 |
"use_cache": false
|
7 |
}
|
model-00001-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:43a1cc3c3c5582d0ec4f21ecc0ec1da402ea2fcdddc83c89ab3c4c4793bed11a
|
3 |
+
size 4927980624
|
model-00002-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f18201c56785c99564e941cb0653d4089c6dd20d3130d8e1f7039f9b65b648c2
|
3 |
+
size 4977500472
|
model-00003-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c1667b3d1e2f67f07ce72a782e6970960dfe8584d74f1720a730e06eda8c1059
|
3 |
+
size 4906314960
|
model-00004-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b5c8da21324e49ea35e7649ba96126f76233cccd8f7b2ee6b0a1230d8b3d599d
|
3 |
+
size 1772618480
|
model.safetensors.index.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
preprocessor_config.json
CHANGED
@@ -1,6 +1,8 @@
|
|
1 |
{
|
2 |
"crop_size": null,
|
|
|
3 |
"default_to_square": true,
|
|
|
4 |
"do_center_crop": null,
|
5 |
"do_convert_rgb": true,
|
6 |
"do_normalize": true,
|
@@ -17,6 +19,7 @@
|
|
17 |
0.26130258,
|
18 |
0.27577711
|
19 |
],
|
|
|
20 |
"max_pixels": 12845056,
|
21 |
"merge_size": 2,
|
22 |
"min_pixels": 3136,
|
@@ -24,6 +27,7 @@
|
|
24 |
"processor_class": "Qwen2_5_VLProcessor",
|
25 |
"resample": 3,
|
26 |
"rescale_factor": 0.00392156862745098,
|
|
|
27 |
"size": {
|
28 |
"longest_edge": 12845056,
|
29 |
"shortest_edge": 3136
|
|
|
1 |
{
|
2 |
"crop_size": null,
|
3 |
+
"data_format": "channels_first",
|
4 |
"default_to_square": true,
|
5 |
+
"device": null,
|
6 |
"do_center_crop": null,
|
7 |
"do_convert_rgb": true,
|
8 |
"do_normalize": true,
|
|
|
19 |
0.26130258,
|
20 |
0.27577711
|
21 |
],
|
22 |
+
"input_data_format": null,
|
23 |
"max_pixels": 12845056,
|
24 |
"merge_size": 2,
|
25 |
"min_pixels": 3136,
|
|
|
27 |
"processor_class": "Qwen2_5_VLProcessor",
|
28 |
"resample": 3,
|
29 |
"rescale_factor": 0.00392156862745098,
|
30 |
+
"return_tensors": null,
|
31 |
"size": {
|
32 |
"longest_edge": 12845056,
|
33 |
"shortest_edge": 3136
|