IMvision12 commited on
Commit
9e53345
·
0 Parent(s):

Super-squash branch 'main' using huggingface_hub

Browse files
.gitattributes ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,90 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ pipeline_tag: image-text-to-text
3
+ license: other
4
+ license_name: deepseek
5
+ license_link: https://huggingface.co/deepseek-ai/deepseek-vl-7b-chat/blob/main/LICENSE
6
+ base_model: deepseek-ai/deepseek-vl-7b-chat
7
+ library_name: kerasformers
8
+ tags:
9
+ - keras
10
+ - kerasformers
11
+ - deepseek-vl
12
+ - vision-language
13
+ - image-text-to-text
14
+ - arxiv:2403.05525
15
+ - pytorch
16
+ - jax
17
+ - tf
18
+ ---
19
+
20
+ ## ***See [our collection](https://huggingface.co/collections/kerasformers/deepseek-vl-6a6ea961fe80d98b7c69b489) for all versions of DeepSeek-VL.***
21
+
22
+ # Run DeepSeek-VL with Keras 3: JAX, PyTorch, or TensorFlow
23
+
24
+ [![GitHub](https://img.shields.io/badge/GitHub-KerasFormers-black?logo=github)](https://github.com/IMvision12/KerasFormers) [![Docs](https://img.shields.io/badge/Docs-DeepSeek--VL--Hybrid-blue)](https://imvision12.github.io/KerasFormers/deepseek_vl_hybrid/) [![Collection](https://img.shields.io/badge/HF-DeepSeek--VL%20collection-yellow)](https://huggingface.co/collections/kerasformers/deepseek-vl-6a6ea961fe80d98b7c69b489)
25
+
26
+ # kerasformers/deepseek_vl_7b_chat
27
+
28
+ Paper: [DeepSeek-VL: Towards Real-World Vision-Language Understanding (arXiv:2403.05525)](https://arxiv.org/abs/2403.05525) · [HF Papers](https://huggingface.co/papers/2403.05525)
29
+
30
+ DeepSeek-VL **7B hybrid** uses a dual vision tower (SigLIP @384 + SAM-style @1024) and a 3-way aligner for finer detail and small text. Import from `kerasformers.models.deepseek_vl_hybrid` (not `deepseek_vl`).
31
+
32
+ For more details on the model, please go to the upstream [model card](https://huggingface.co/deepseek-ai/deepseek-vl-7b-chat).
33
+
34
+ Pure-**Keras 3** conversion of [`deepseek-ai/deepseek-vl-7b-chat`](https://huggingface.co/deepseek-ai/deepseek-vl-7b-chat) for [kerasformers](https://github.com/IMvision12/KerasFormers). One implementation runs unmodified on **TensorFlow / Torch / JAX**.
35
+
36
+ This is a **vision-language** (chat) checkpoint (`DeepseekVLHybridConditionalGenerate`, 7B chat (SigLIP + SAM hybrid)).
37
+
38
+ ## ✨ Quick start
39
+
40
+ ```python
41
+ import os
42
+ os.environ["KERAS_BACKEND"] = "torch" # or "jax" / "tensorflow"
43
+
44
+ from PIL import Image
45
+ from kerasformers.models.deepseek_vl_hybrid import (
46
+ DeepseekVLHybridConditionalGenerate,
47
+ DeepseekVLHybridProcessor,
48
+ )
49
+
50
+ model = DeepseekVLHybridConditionalGenerate.from_weights("kerasformers/deepseek_vl_7b_chat")
51
+ processor = DeepseekVLHybridProcessor.from_weights("kerasformers/deepseek_vl_7b_chat")
52
+
53
+ image = Image.open("your_image.jpg")
54
+ inputs = processor(
55
+ conversation=[
56
+ {
57
+ "role": "user",
58
+ "content": [
59
+ {"type": "image", "image": image},
60
+ {"type": "text", "text": "Describe this image in one sentence."},
61
+ ],
62
+ }
63
+ ]
64
+ )
65
+ outputs = model.generate(**inputs, max_new_tokens=64)
66
+ print(processor.decode(outputs[0]))
67
+ ```
68
+
69
+ Load any DeepSeek-VL variant the same way with `from_weights("kerasformers/<variant>")`:
70
+
71
+ | Variant | Hub | Notes |
72
+ |---|---|---|
73
+ | `deepseek_vl_1.3b_base` | [`kerasformers/deepseek_vl_1.3b_base`](https://huggingface.co/kerasformers/deepseek_vl_1.3b_base) | 1.3B base |
74
+ | `deepseek_vl_1.3b_chat` | [`kerasformers/deepseek_vl_1.3b_chat`](https://huggingface.co/kerasformers/deepseek_vl_1.3b_chat) | 1.3B chat |
75
+ | `deepseek_vl_7b_base` | [`kerasformers/deepseek_vl_7b_base`](https://huggingface.co/kerasformers/deepseek_vl_7b_base) | 7B hybrid base |
76
+ | `deepseek_vl_7b_chat` | [`kerasformers/deepseek_vl_7b_chat`](https://huggingface.co/kerasformers/deepseek_vl_7b_chat) | 7B hybrid chat |
77
+
78
+ ## Tips
79
+
80
+ - Set `KERAS_BACKEND` **before** importing Keras / kerasformers.
81
+ - Prefer `Processor.from_weights(...)` so vision + tokenizer match.
82
+ - 7B hybrid is a **different package** than 1.3B; do not mix imports.
83
+ - See [docs](https://imvision12.github.io/KerasFormers/deepseek_vl_hybrid/) and [Loading Weights](https://imvision12.github.io/KerasFormers/loading_weights/).
84
+ - Community / upstream safetensors still work via the `hf:` prefix, e.g. `DeepseekVLHybridConditionalGenerate.from_weights("hf:deepseek-ai/deepseek-vl-7b-chat")`.
85
+
86
+ ## Special Thanks
87
+
88
+ A huge thank you to the DeepSeek-VL authors for creating and releasing these models.
89
+
90
+ License: DeepSeek (`other` / deepseek). See [LICENSE](https://huggingface.co/deepseek-ai/deepseek-vl-7b-chat/blob/main/LICENSE).
kf_config.json ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "library_name": "kerasformers",
3
+ "kerasformers_version": "1.2.1",
4
+ "model_module": "kerasformers.models.deepseek_vl_hybrid",
5
+ "model_class": "DeepseekVLHybridModel",
6
+ "variant": "deepseek_vl_7b_chat",
7
+ "weights": "model.weights.json",
8
+ "schema_version": 2,
9
+ "weight_dtype": "float16",
10
+ "model_type": "deepseek_vl_hybrid",
11
+ "text_config": {
12
+ "vocab_size": 102400,
13
+ "embed_dim": 4096,
14
+ "mlp_dim": 11008,
15
+ "num_layers": 30,
16
+ "num_heads": 32,
17
+ "num_kv_heads": 32,
18
+ "head_dim": 128,
19
+ "norm_eps": 1e-06,
20
+ "rope_theta": 10000.0,
21
+ "tie_embeddings": false
22
+ },
23
+ "vision_config": {
24
+ "embed_dim": 1024,
25
+ "mlp_dim": 4096,
26
+ "num_layers": 24,
27
+ "num_heads": 16,
28
+ "norm_eps": 1e-06,
29
+ "image_size": 384,
30
+ "patch_size": 16
31
+ },
32
+ "high_res_config": {
33
+ "embed_dim": 768,
34
+ "mlp_dim": 3072,
35
+ "num_layers": 12,
36
+ "num_heads": 12,
37
+ "image_size": 1024,
38
+ "patch_size": 16,
39
+ "output_channels": 256,
40
+ "window_size": 14,
41
+ "global_attn_indexes": [
42
+ 2,
43
+ 5,
44
+ 8,
45
+ 11
46
+ ],
47
+ "norm_eps": 1e-06
48
+ },
49
+ "image_token_id": 100015
50
+ }
kf_preprocessor.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "library_name": "kerasformers",
3
+ "kerasformers_version": "1.1.3",
4
+ "preprocessor_module": "kerasformers.models.deepseek_vl_hybrid",
5
+ "preprocessor_class": "DeepseekVLHybridImageProcessor",
6
+ "variant": "deepseek_vl_7b_chat",
7
+ "size": 384,
8
+ "high_res_size": 1024,
9
+ "min_size": 14,
10
+ "image_mean": [
11
+ 0.5,
12
+ 0.5,
13
+ 0.5
14
+ ],
15
+ "image_std": [
16
+ 0.5,
17
+ 0.5,
18
+ 0.5
19
+ ],
20
+ "high_res_image_mean": [
21
+ 0.48145466,
22
+ 0.4578275,
23
+ 0.40821073
24
+ ],
25
+ "high_res_image_std": [
26
+ 0.26862954,
27
+ 0.26130258,
28
+ 0.27577711
29
+ ],
30
+ "background_color": [
31
+ 127,
32
+ 127,
33
+ 127
34
+ ],
35
+ "high_res_background_color": [
36
+ 122,
37
+ 116,
38
+ 104
39
+ ],
40
+ "data_format": null
41
+ }
model.weights.json ADDED
@@ -0,0 +1,1148 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 13823920652.0
4
+ },
5
+ "weight_map": {
6
+ "/vars": "model_00000.weights.h5",
7
+ "/layers/embedding/vars": "model_00000.weights.h5",
8
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/attention/key/vars": [
9
+ "model_00000.weights.h5"
10
+ ],
11
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/attention/output_proj/vars": [
12
+ "model_00000.weights.h5"
13
+ ],
14
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/attention/query/vars": [
15
+ "model_00000.weights.h5"
16
+ ],
17
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/attention/value/vars": [
18
+ "model_00000.weights.h5"
19
+ ],
20
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/fc1/vars": [
21
+ "model_00000.weights.h5"
22
+ ],
23
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/fc2/vars": [
24
+ "model_00000.weights.h5"
25
+ ],
26
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/layer_norm1/vars": [
27
+ "model_00000.weights.h5"
28
+ ],
29
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer/layer_norm2/vars": [
30
+ "model_00000.weights.h5"
31
+ ],
32
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/attention/key/vars": [
33
+ "model_00000.weights.h5"
34
+ ],
35
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/attention/output_proj/vars": [
36
+ "model_00000.weights.h5"
37
+ ],
38
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/attention/query/vars": [
39
+ "model_00000.weights.h5"
40
+ ],
41
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/attention/value/vars": [
42
+ "model_00000.weights.h5"
43
+ ],
44
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/fc1/vars": [
45
+ "model_00000.weights.h5"
46
+ ],
47
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/fc2/vars": [
48
+ "model_00000.weights.h5"
49
+ ],
50
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/layer_norm1/vars": [
51
+ "model_00000.weights.h5"
52
+ ],
53
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_1/layer_norm2/vars": [
54
+ "model_00000.weights.h5"
55
+ ],
56
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/attention/key/vars": [
57
+ "model_00000.weights.h5"
58
+ ],
59
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/attention/output_proj/vars": [
60
+ "model_00000.weights.h5"
61
+ ],
62
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/attention/query/vars": [
63
+ "model_00000.weights.h5"
64
+ ],
65
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/attention/value/vars": [
66
+ "model_00000.weights.h5"
67
+ ],
68
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/fc1/vars": [
69
+ "model_00000.weights.h5"
70
+ ],
71
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/fc2/vars": [
72
+ "model_00000.weights.h5"
73
+ ],
74
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/layer_norm1/vars": [
75
+ "model_00000.weights.h5"
76
+ ],
77
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_2/layer_norm2/vars": [
78
+ "model_00000.weights.h5"
79
+ ],
80
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/attention/key/vars": [
81
+ "model_00000.weights.h5"
82
+ ],
83
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/attention/output_proj/vars": [
84
+ "model_00000.weights.h5"
85
+ ],
86
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/attention/query/vars": [
87
+ "model_00000.weights.h5"
88
+ ],
89
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/attention/value/vars": [
90
+ "model_00000.weights.h5"
91
+ ],
92
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/fc1/vars": [
93
+ "model_00000.weights.h5"
94
+ ],
95
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/fc2/vars": [
96
+ "model_00000.weights.h5"
97
+ ],
98
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/layer_norm1/vars": [
99
+ "model_00000.weights.h5"
100
+ ],
101
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_3/layer_norm2/vars": [
102
+ "model_00000.weights.h5"
103
+ ],
104
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/attention/key/vars": [
105
+ "model_00000.weights.h5"
106
+ ],
107
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/attention/output_proj/vars": [
108
+ "model_00000.weights.h5"
109
+ ],
110
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/attention/query/vars": [
111
+ "model_00000.weights.h5"
112
+ ],
113
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/attention/value/vars": [
114
+ "model_00000.weights.h5"
115
+ ],
116
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/fc1/vars": [
117
+ "model_00000.weights.h5"
118
+ ],
119
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/fc2/vars": [
120
+ "model_00000.weights.h5"
121
+ ],
122
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/layer_norm1/vars": [
123
+ "model_00000.weights.h5"
124
+ ],
125
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_4/layer_norm2/vars": [
126
+ "model_00000.weights.h5"
127
+ ],
128
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/attention/key/vars": [
129
+ "model_00000.weights.h5"
130
+ ],
131
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/attention/output_proj/vars": [
132
+ "model_00000.weights.h5"
133
+ ],
134
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/attention/query/vars": [
135
+ "model_00000.weights.h5"
136
+ ],
137
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/attention/value/vars": [
138
+ "model_00000.weights.h5"
139
+ ],
140
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/fc1/vars": [
141
+ "model_00000.weights.h5"
142
+ ],
143
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/fc2/vars": [
144
+ "model_00000.weights.h5"
145
+ ],
146
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/layer_norm1/vars": [
147
+ "model_00000.weights.h5"
148
+ ],
149
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_5/layer_norm2/vars": [
150
+ "model_00000.weights.h5"
151
+ ],
152
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/attention/key/vars": [
153
+ "model_00000.weights.h5"
154
+ ],
155
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/attention/output_proj/vars": [
156
+ "model_00000.weights.h5"
157
+ ],
158
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/attention/query/vars": [
159
+ "model_00000.weights.h5"
160
+ ],
161
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/attention/value/vars": [
162
+ "model_00000.weights.h5"
163
+ ],
164
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/fc1/vars": [
165
+ "model_00000.weights.h5"
166
+ ],
167
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/fc2/vars": [
168
+ "model_00000.weights.h5"
169
+ ],
170
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/layer_norm1/vars": [
171
+ "model_00000.weights.h5"
172
+ ],
173
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_6/layer_norm2/vars": [
174
+ "model_00000.weights.h5"
175
+ ],
176
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/attention/key/vars": [
177
+ "model_00000.weights.h5"
178
+ ],
179
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/attention/output_proj/vars": [
180
+ "model_00000.weights.h5"
181
+ ],
182
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/attention/query/vars": [
183
+ "model_00000.weights.h5"
184
+ ],
185
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/attention/value/vars": [
186
+ "model_00000.weights.h5"
187
+ ],
188
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/fc1/vars": [
189
+ "model_00000.weights.h5"
190
+ ],
191
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/fc2/vars": [
192
+ "model_00000.weights.h5"
193
+ ],
194
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/layer_norm1/vars": [
195
+ "model_00000.weights.h5"
196
+ ],
197
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_7/layer_norm2/vars": [
198
+ "model_00000.weights.h5"
199
+ ],
200
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/attention/key/vars": [
201
+ "model_00000.weights.h5"
202
+ ],
203
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/attention/output_proj/vars": [
204
+ "model_00000.weights.h5"
205
+ ],
206
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/attention/query/vars": [
207
+ "model_00000.weights.h5"
208
+ ],
209
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/attention/value/vars": [
210
+ "model_00000.weights.h5"
211
+ ],
212
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/fc1/vars": [
213
+ "model_00000.weights.h5"
214
+ ],
215
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/fc2/vars": [
216
+ "model_00000.weights.h5"
217
+ ],
218
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/layer_norm1/vars": [
219
+ "model_00000.weights.h5"
220
+ ],
221
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_8/layer_norm2/vars": [
222
+ "model_00000.weights.h5"
223
+ ],
224
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/attention/key/vars": [
225
+ "model_00000.weights.h5"
226
+ ],
227
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/attention/output_proj/vars": [
228
+ "model_00000.weights.h5"
229
+ ],
230
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/attention/query/vars": [
231
+ "model_00000.weights.h5"
232
+ ],
233
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/attention/value/vars": [
234
+ "model_00000.weights.h5"
235
+ ],
236
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/fc1/vars": [
237
+ "model_00000.weights.h5"
238
+ ],
239
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/fc2/vars": [
240
+ "model_00000.weights.h5"
241
+ ],
242
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/layer_norm1/vars": [
243
+ "model_00000.weights.h5"
244
+ ],
245
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_9/layer_norm2/vars": [
246
+ "model_00000.weights.h5"
247
+ ],
248
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/attention/key/vars": [
249
+ "model_00000.weights.h5"
250
+ ],
251
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/attention/output_proj/vars": [
252
+ "model_00000.weights.h5"
253
+ ],
254
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/attention/query/vars": [
255
+ "model_00000.weights.h5"
256
+ ],
257
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/attention/value/vars": [
258
+ "model_00000.weights.h5"
259
+ ],
260
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/fc1/vars": [
261
+ "model_00000.weights.h5"
262
+ ],
263
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/fc2/vars": [
264
+ "model_00000.weights.h5"
265
+ ],
266
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/layer_norm1/vars": [
267
+ "model_00000.weights.h5"
268
+ ],
269
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_10/layer_norm2/vars": [
270
+ "model_00000.weights.h5"
271
+ ],
272
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/attention/key/vars": [
273
+ "model_00000.weights.h5"
274
+ ],
275
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/attention/output_proj/vars": [
276
+ "model_00000.weights.h5"
277
+ ],
278
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/attention/query/vars": [
279
+ "model_00000.weights.h5"
280
+ ],
281
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/attention/value/vars": [
282
+ "model_00000.weights.h5"
283
+ ],
284
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/fc1/vars": [
285
+ "model_00000.weights.h5"
286
+ ],
287
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/fc2/vars": [
288
+ "model_00000.weights.h5"
289
+ ],
290
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/layer_norm1/vars": [
291
+ "model_00000.weights.h5"
292
+ ],
293
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_11/layer_norm2/vars": [
294
+ "model_00000.weights.h5"
295
+ ],
296
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/attention/key/vars": [
297
+ "model_00000.weights.h5"
298
+ ],
299
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/attention/output_proj/vars": [
300
+ "model_00000.weights.h5"
301
+ ],
302
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/attention/query/vars": [
303
+ "model_00000.weights.h5"
304
+ ],
305
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/attention/value/vars": [
306
+ "model_00000.weights.h5"
307
+ ],
308
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/fc1/vars": [
309
+ "model_00000.weights.h5"
310
+ ],
311
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/fc2/vars": [
312
+ "model_00000.weights.h5"
313
+ ],
314
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/layer_norm1/vars": [
315
+ "model_00000.weights.h5"
316
+ ],
317
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_12/layer_norm2/vars": [
318
+ "model_00000.weights.h5"
319
+ ],
320
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/attention/key/vars": [
321
+ "model_00000.weights.h5"
322
+ ],
323
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/attention/output_proj/vars": [
324
+ "model_00000.weights.h5"
325
+ ],
326
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/attention/query/vars": [
327
+ "model_00000.weights.h5"
328
+ ],
329
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/attention/value/vars": [
330
+ "model_00000.weights.h5"
331
+ ],
332
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/fc1/vars": [
333
+ "model_00000.weights.h5"
334
+ ],
335
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/fc2/vars": [
336
+ "model_00000.weights.h5"
337
+ ],
338
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/layer_norm1/vars": [
339
+ "model_00000.weights.h5"
340
+ ],
341
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_13/layer_norm2/vars": [
342
+ "model_00000.weights.h5"
343
+ ],
344
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/attention/key/vars": [
345
+ "model_00000.weights.h5"
346
+ ],
347
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/attention/output_proj/vars": [
348
+ "model_00000.weights.h5"
349
+ ],
350
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/attention/query/vars": [
351
+ "model_00000.weights.h5"
352
+ ],
353
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/attention/value/vars": [
354
+ "model_00000.weights.h5"
355
+ ],
356
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/fc1/vars": [
357
+ "model_00000.weights.h5"
358
+ ],
359
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/fc2/vars": [
360
+ "model_00000.weights.h5"
361
+ ],
362
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/layer_norm1/vars": [
363
+ "model_00000.weights.h5"
364
+ ],
365
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_14/layer_norm2/vars": [
366
+ "model_00000.weights.h5"
367
+ ],
368
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/attention/key/vars": [
369
+ "model_00000.weights.h5"
370
+ ],
371
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/attention/output_proj/vars": [
372
+ "model_00000.weights.h5"
373
+ ],
374
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/attention/query/vars": [
375
+ "model_00000.weights.h5"
376
+ ],
377
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/attention/value/vars": [
378
+ "model_00000.weights.h5"
379
+ ],
380
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/fc1/vars": [
381
+ "model_00000.weights.h5"
382
+ ],
383
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/fc2/vars": [
384
+ "model_00000.weights.h5"
385
+ ],
386
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/layer_norm1/vars": [
387
+ "model_00000.weights.h5"
388
+ ],
389
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_15/layer_norm2/vars": [
390
+ "model_00000.weights.h5"
391
+ ],
392
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/attention/key/vars": [
393
+ "model_00000.weights.h5"
394
+ ],
395
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/attention/output_proj/vars": [
396
+ "model_00000.weights.h5"
397
+ ],
398
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/attention/query/vars": [
399
+ "model_00000.weights.h5"
400
+ ],
401
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/attention/value/vars": [
402
+ "model_00000.weights.h5"
403
+ ],
404
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/fc1/vars": [
405
+ "model_00000.weights.h5"
406
+ ],
407
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/fc2/vars": [
408
+ "model_00000.weights.h5"
409
+ ],
410
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/layer_norm1/vars": [
411
+ "model_00000.weights.h5"
412
+ ],
413
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_16/layer_norm2/vars": [
414
+ "model_00000.weights.h5"
415
+ ],
416
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/attention/key/vars": [
417
+ "model_00000.weights.h5"
418
+ ],
419
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/attention/output_proj/vars": [
420
+ "model_00000.weights.h5"
421
+ ],
422
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/attention/query/vars": [
423
+ "model_00000.weights.h5"
424
+ ],
425
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/attention/value/vars": [
426
+ "model_00000.weights.h5"
427
+ ],
428
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/fc1/vars": [
429
+ "model_00000.weights.h5"
430
+ ],
431
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/fc2/vars": [
432
+ "model_00000.weights.h5"
433
+ ],
434
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/layer_norm1/vars": [
435
+ "model_00000.weights.h5"
436
+ ],
437
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_17/layer_norm2/vars": [
438
+ "model_00000.weights.h5"
439
+ ],
440
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/attention/key/vars": [
441
+ "model_00000.weights.h5"
442
+ ],
443
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/attention/output_proj/vars": [
444
+ "model_00000.weights.h5"
445
+ ],
446
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/attention/query/vars": [
447
+ "model_00000.weights.h5"
448
+ ],
449
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/attention/value/vars": [
450
+ "model_00000.weights.h5"
451
+ ],
452
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/fc1/vars": [
453
+ "model_00000.weights.h5"
454
+ ],
455
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/fc2/vars": [
456
+ "model_00000.weights.h5"
457
+ ],
458
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/layer_norm1/vars": [
459
+ "model_00000.weights.h5"
460
+ ],
461
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_18/layer_norm2/vars": [
462
+ "model_00000.weights.h5"
463
+ ],
464
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/attention/key/vars": [
465
+ "model_00000.weights.h5"
466
+ ],
467
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/attention/output_proj/vars": [
468
+ "model_00000.weights.h5"
469
+ ],
470
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/attention/query/vars": [
471
+ "model_00000.weights.h5"
472
+ ],
473
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/attention/value/vars": [
474
+ "model_00000.weights.h5"
475
+ ],
476
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/fc1/vars": [
477
+ "model_00000.weights.h5"
478
+ ],
479
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/fc2/vars": [
480
+ "model_00000.weights.h5"
481
+ ],
482
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/layer_norm1/vars": [
483
+ "model_00000.weights.h5"
484
+ ],
485
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_19/layer_norm2/vars": [
486
+ "model_00000.weights.h5"
487
+ ],
488
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/attention/key/vars": [
489
+ "model_00000.weights.h5"
490
+ ],
491
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/attention/output_proj/vars": [
492
+ "model_00000.weights.h5"
493
+ ],
494
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/attention/query/vars": [
495
+ "model_00000.weights.h5"
496
+ ],
497
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/attention/value/vars": [
498
+ "model_00000.weights.h5"
499
+ ],
500
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/fc1/vars": [
501
+ "model_00000.weights.h5"
502
+ ],
503
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/fc2/vars": [
504
+ "model_00000.weights.h5"
505
+ ],
506
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/layer_norm1/vars": [
507
+ "model_00000.weights.h5"
508
+ ],
509
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_20/layer_norm2/vars": [
510
+ "model_00000.weights.h5"
511
+ ],
512
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/attention/key/vars": [
513
+ "model_00000.weights.h5"
514
+ ],
515
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/attention/output_proj/vars": [
516
+ "model_00000.weights.h5"
517
+ ],
518
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/attention/query/vars": [
519
+ "model_00000.weights.h5"
520
+ ],
521
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/attention/value/vars": [
522
+ "model_00000.weights.h5"
523
+ ],
524
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/fc1/vars": [
525
+ "model_00000.weights.h5"
526
+ ],
527
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/fc2/vars": [
528
+ "model_00000.weights.h5"
529
+ ],
530
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/layer_norm1/vars": [
531
+ "model_00000.weights.h5"
532
+ ],
533
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_21/layer_norm2/vars": [
534
+ "model_00000.weights.h5"
535
+ ],
536
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/attention/key/vars": [
537
+ "model_00000.weights.h5"
538
+ ],
539
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/attention/output_proj/vars": [
540
+ "model_00000.weights.h5"
541
+ ],
542
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/attention/query/vars": [
543
+ "model_00000.weights.h5"
544
+ ],
545
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/attention/value/vars": [
546
+ "model_00000.weights.h5"
547
+ ],
548
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/fc1/vars": [
549
+ "model_00000.weights.h5"
550
+ ],
551
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/fc2/vars": [
552
+ "model_00000.weights.h5"
553
+ ],
554
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/layer_norm1/vars": [
555
+ "model_00000.weights.h5"
556
+ ],
557
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_22/layer_norm2/vars": [
558
+ "model_00000.weights.h5"
559
+ ],
560
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/attention/key/vars": [
561
+ "model_00000.weights.h5"
562
+ ],
563
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/attention/output_proj/vars": [
564
+ "model_00000.weights.h5"
565
+ ],
566
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/attention/query/vars": [
567
+ "model_00000.weights.h5"
568
+ ],
569
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/attention/value/vars": [
570
+ "model_00000.weights.h5"
571
+ ],
572
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/fc1/vars": [
573
+ "model_00000.weights.h5"
574
+ ],
575
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/fc2/vars": [
576
+ "model_00000.weights.h5"
577
+ ],
578
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/layer_norm1/vars": [
579
+ "model_00000.weights.h5"
580
+ ],
581
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/blocks/deepseek_vl_vision_layer_23/layer_norm2/vars": [
582
+ "model_00000.weights.h5"
583
+ ],
584
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/patch_embed/vars": [
585
+ "model_00000.weights.h5"
586
+ ],
587
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/position_embedding/vars": "model_00000.weights.h5",
588
+ "/layers/hybrid_image_features/refs/deepseek_vl_vision_model/post_layernorm/vars": [
589
+ "model_00000.weights.h5"
590
+ ],
591
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer/attn/vars": [
592
+ "model_00000.weights.h5"
593
+ ],
594
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer/attn/proj/vars": [
595
+ "model_00000.weights.h5"
596
+ ],
597
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer/attn/qkv/vars": [
598
+ "model_00000.weights.h5"
599
+ ],
600
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer/layer_norm1/vars": [
601
+ "model_00000.weights.h5"
602
+ ],
603
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer/layer_norm2/vars": [
604
+ "model_00000.weights.h5"
605
+ ],
606
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer/mlp_lin1/vars": [
607
+ "model_00000.weights.h5"
608
+ ],
609
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer/mlp_lin2/vars": [
610
+ "model_00000.weights.h5"
611
+ ],
612
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_1/attn/vars": [
613
+ "model_00000.weights.h5"
614
+ ],
615
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_1/attn/proj/vars": [
616
+ "model_00000.weights.h5"
617
+ ],
618
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_1/attn/qkv/vars": [
619
+ "model_00000.weights.h5"
620
+ ],
621
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_1/layer_norm1/vars": [
622
+ "model_00000.weights.h5"
623
+ ],
624
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_1/layer_norm2/vars": [
625
+ "model_00000.weights.h5"
626
+ ],
627
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_1/mlp_lin1/vars": [
628
+ "model_00000.weights.h5"
629
+ ],
630
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_1/mlp_lin2/vars": [
631
+ "model_00000.weights.h5"
632
+ ],
633
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_2/attn/vars": [
634
+ "model_00000.weights.h5"
635
+ ],
636
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_2/attn/proj/vars": [
637
+ "model_00000.weights.h5"
638
+ ],
639
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_2/attn/qkv/vars": [
640
+ "model_00000.weights.h5"
641
+ ],
642
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_2/layer_norm1/vars": [
643
+ "model_00000.weights.h5"
644
+ ],
645
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_2/layer_norm2/vars": [
646
+ "model_00000.weights.h5"
647
+ ],
648
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_2/mlp_lin1/vars": [
649
+ "model_00000.weights.h5"
650
+ ],
651
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_2/mlp_lin2/vars": [
652
+ "model_00000.weights.h5"
653
+ ],
654
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_3/attn/vars": [
655
+ "model_00000.weights.h5"
656
+ ],
657
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_3/attn/proj/vars": [
658
+ "model_00000.weights.h5"
659
+ ],
660
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_3/attn/qkv/vars": [
661
+ "model_00000.weights.h5"
662
+ ],
663
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_3/layer_norm1/vars": [
664
+ "model_00000.weights.h5"
665
+ ],
666
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_3/layer_norm2/vars": [
667
+ "model_00000.weights.h5"
668
+ ],
669
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_3/mlp_lin1/vars": [
670
+ "model_00000.weights.h5"
671
+ ],
672
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_3/mlp_lin2/vars": [
673
+ "model_00000.weights.h5"
674
+ ],
675
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_4/attn/vars": [
676
+ "model_00000.weights.h5"
677
+ ],
678
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_4/attn/proj/vars": [
679
+ "model_00000.weights.h5"
680
+ ],
681
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_4/attn/qkv/vars": [
682
+ "model_00000.weights.h5"
683
+ ],
684
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_4/layer_norm1/vars": [
685
+ "model_00000.weights.h5"
686
+ ],
687
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_4/layer_norm2/vars": [
688
+ "model_00000.weights.h5"
689
+ ],
690
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_4/mlp_lin1/vars": [
691
+ "model_00000.weights.h5"
692
+ ],
693
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_4/mlp_lin2/vars": [
694
+ "model_00000.weights.h5"
695
+ ],
696
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_5/attn/vars": [
697
+ "model_00000.weights.h5"
698
+ ],
699
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_5/attn/proj/vars": [
700
+ "model_00000.weights.h5"
701
+ ],
702
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_5/attn/qkv/vars": [
703
+ "model_00000.weights.h5"
704
+ ],
705
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_5/layer_norm1/vars": [
706
+ "model_00000.weights.h5"
707
+ ],
708
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_5/layer_norm2/vars": [
709
+ "model_00000.weights.h5"
710
+ ],
711
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_5/mlp_lin1/vars": [
712
+ "model_00000.weights.h5"
713
+ ],
714
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_5/mlp_lin2/vars": [
715
+ "model_00000.weights.h5"
716
+ ],
717
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_6/attn/vars": [
718
+ "model_00000.weights.h5"
719
+ ],
720
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_6/attn/proj/vars": [
721
+ "model_00000.weights.h5"
722
+ ],
723
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_6/attn/qkv/vars": [
724
+ "model_00000.weights.h5"
725
+ ],
726
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_6/layer_norm1/vars": [
727
+ "model_00000.weights.h5"
728
+ ],
729
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_6/layer_norm2/vars": [
730
+ "model_00000.weights.h5"
731
+ ],
732
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_6/mlp_lin1/vars": [
733
+ "model_00000.weights.h5"
734
+ ],
735
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_6/mlp_lin2/vars": [
736
+ "model_00000.weights.h5"
737
+ ],
738
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_7/attn/vars": [
739
+ "model_00000.weights.h5"
740
+ ],
741
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_7/attn/proj/vars": [
742
+ "model_00000.weights.h5"
743
+ ],
744
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_7/attn/qkv/vars": [
745
+ "model_00000.weights.h5"
746
+ ],
747
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_7/layer_norm1/vars": [
748
+ "model_00000.weights.h5"
749
+ ],
750
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_7/layer_norm2/vars": [
751
+ "model_00000.weights.h5"
752
+ ],
753
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_7/mlp_lin1/vars": [
754
+ "model_00000.weights.h5"
755
+ ],
756
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_7/mlp_lin2/vars": [
757
+ "model_00000.weights.h5"
758
+ ],
759
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_8/attn/vars": [
760
+ "model_00000.weights.h5"
761
+ ],
762
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_8/attn/proj/vars": [
763
+ "model_00000.weights.h5"
764
+ ],
765
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_8/attn/qkv/vars": [
766
+ "model_00000.weights.h5"
767
+ ],
768
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_8/layer_norm1/vars": [
769
+ "model_00000.weights.h5"
770
+ ],
771
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_8/layer_norm2/vars": [
772
+ "model_00000.weights.h5"
773
+ ],
774
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_8/mlp_lin1/vars": [
775
+ "model_00000.weights.h5"
776
+ ],
777
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_8/mlp_lin2/vars": [
778
+ "model_00000.weights.h5"
779
+ ],
780
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_9/attn/vars": [
781
+ "model_00000.weights.h5"
782
+ ],
783
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_9/attn/proj/vars": [
784
+ "model_00000.weights.h5"
785
+ ],
786
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_9/attn/qkv/vars": [
787
+ "model_00000.weights.h5"
788
+ ],
789
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_9/layer_norm1/vars": [
790
+ "model_00000.weights.h5"
791
+ ],
792
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_9/layer_norm2/vars": [
793
+ "model_00000.weights.h5"
794
+ ],
795
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_9/mlp_lin1/vars": [
796
+ "model_00000.weights.h5"
797
+ ],
798
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_9/mlp_lin2/vars": [
799
+ "model_00000.weights.h5"
800
+ ],
801
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_10/attn/vars": [
802
+ "model_00000.weights.h5"
803
+ ],
804
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_10/attn/proj/vars": [
805
+ "model_00000.weights.h5"
806
+ ],
807
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_10/attn/qkv/vars": [
808
+ "model_00000.weights.h5"
809
+ ],
810
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_10/layer_norm1/vars": [
811
+ "model_00000.weights.h5"
812
+ ],
813
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_10/layer_norm2/vars": [
814
+ "model_00000.weights.h5"
815
+ ],
816
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_10/mlp_lin1/vars": [
817
+ "model_00000.weights.h5"
818
+ ],
819
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_10/mlp_lin2/vars": [
820
+ "model_00000.weights.h5"
821
+ ],
822
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_11/attn/vars": [
823
+ "model_00000.weights.h5"
824
+ ],
825
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_11/attn/proj/vars": [
826
+ "model_00000.weights.h5"
827
+ ],
828
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_11/attn/qkv/vars": [
829
+ "model_00000.weights.h5"
830
+ ],
831
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_11/layer_norm1/vars": [
832
+ "model_00000.weights.h5"
833
+ ],
834
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_11/layer_norm2/vars": [
835
+ "model_00000.weights.h5"
836
+ ],
837
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_11/mlp_lin1/vars": [
838
+ "model_00000.weights.h5"
839
+ ],
840
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/blocks/sam_vision_layer_11/mlp_lin2/vars": [
841
+ "model_00000.weights.h5"
842
+ ],
843
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/neck_conv1/vars": "model_00000.weights.h5",
844
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/neck_conv2/vars": "model_00000.weights.h5",
845
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/neck_ln1/vars": [
846
+ "model_00000.weights.h5"
847
+ ],
848
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/neck_ln2/vars": [
849
+ "model_00000.weights.h5"
850
+ ],
851
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/patch_embed/vars": [
852
+ "model_00000.weights.h5"
853
+ ],
854
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_sam_encoder/pos_embed/vars": [
855
+ "model_00000.weights.h5"
856
+ ],
857
+ "/layers/hybrid_image_features/refs/deepseek_vl_sam_vision_neck/conv1/vars": "model_00000.weights.h5",
858
+ "/layers/hybrid_image_features/refs/deepseek_vl_sam_vision_neck/conv2/vars": "model_00000.weights.h5",
859
+ "/layers/hybrid_image_features/refs/deepseek_vl_sam_vision_neck/layer_norm1/vars": [
860
+ "model_00000.weights.h5"
861
+ ],
862
+ "/layers/hybrid_image_features/refs/deepseek_vl_sam_vision_neck/layer_norm2/vars": [
863
+ "model_00000.weights.h5"
864
+ ],
865
+ "/layers/hybrid_image_features/refs/deepseek_vl_sam_vision_proj/conv1/vars": "model_00000.weights.h5",
866
+ "/layers/hybrid_image_features/refs/deepseek_vl_sam_vision_proj/conv2/vars": "model_00000.weights.h5",
867
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_aligner/high_res_vision_proj/vars": [
868
+ "model_00000.weights.h5"
869
+ ],
870
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_aligner/proj/vars": [
871
+ "model_00000.weights.h5"
872
+ ],
873
+ "/layers/hybrid_image_features/refs/deepseek_vl_hybrid_aligner/vision_proj/vars": [
874
+ "model_00000.weights.h5"
875
+ ],
876
+ "/layers/deepseek_vl_text_decoder_layer/attention/key/vars": "model_00000.weights.h5",
877
+ "/layers/deepseek_vl_text_decoder_layer/attention/output_proj/vars": "model_00000.weights.h5",
878
+ "/layers/deepseek_vl_text_decoder_layer/attention/query/vars": "model_00000.weights.h5",
879
+ "/layers/deepseek_vl_text_decoder_layer/attention/value/vars": "model_00000.weights.h5",
880
+ "/layers/deepseek_vl_text_decoder_layer/attention_norm/vars": "model_00000.weights.h5",
881
+ "/layers/deepseek_vl_text_decoder_layer/mlp/down/vars": "model_00000.weights.h5",
882
+ "/layers/deepseek_vl_text_decoder_layer/mlp/gate/vars": "model_00000.weights.h5",
883
+ "/layers/deepseek_vl_text_decoder_layer/mlp/up/vars": "model_00000.weights.h5",
884
+ "/layers/deepseek_vl_text_decoder_layer/mlp_norm/vars": "model_00000.weights.h5",
885
+ "/layers/deepseek_vl_text_decoder_layer_1/attention/key/vars": "model_00000.weights.h5",
886
+ "/layers/deepseek_vl_text_decoder_layer_1/attention/output_proj/vars": "model_00000.weights.h5",
887
+ "/layers/deepseek_vl_text_decoder_layer_1/attention/query/vars": "model_00000.weights.h5",
888
+ "/layers/deepseek_vl_text_decoder_layer_1/attention/value/vars": "model_00000.weights.h5",
889
+ "/layers/deepseek_vl_text_decoder_layer_1/attention_norm/vars": "model_00000.weights.h5",
890
+ "/layers/deepseek_vl_text_decoder_layer_1/mlp/down/vars": "model_00000.weights.h5",
891
+ "/layers/deepseek_vl_text_decoder_layer_1/mlp/gate/vars": "model_00000.weights.h5",
892
+ "/layers/deepseek_vl_text_decoder_layer_1/mlp/up/vars": "model_00000.weights.h5",
893
+ "/layers/deepseek_vl_text_decoder_layer_1/mlp_norm/vars": "model_00000.weights.h5",
894
+ "/layers/deepseek_vl_text_decoder_layer_2/attention/key/vars": "model_00000.weights.h5",
895
+ "/layers/deepseek_vl_text_decoder_layer_2/attention/output_proj/vars": "model_00000.weights.h5",
896
+ "/layers/deepseek_vl_text_decoder_layer_2/attention/query/vars": "model_00000.weights.h5",
897
+ "/layers/deepseek_vl_text_decoder_layer_2/attention/value/vars": "model_00000.weights.h5",
898
+ "/layers/deepseek_vl_text_decoder_layer_2/attention_norm/vars": "model_00000.weights.h5",
899
+ "/layers/deepseek_vl_text_decoder_layer_2/mlp/down/vars": "model_00000.weights.h5",
900
+ "/layers/deepseek_vl_text_decoder_layer_2/mlp/gate/vars": "model_00000.weights.h5",
901
+ "/layers/deepseek_vl_text_decoder_layer_2/mlp/up/vars": "model_00000.weights.h5",
902
+ "/layers/deepseek_vl_text_decoder_layer_2/mlp_norm/vars": "model_00000.weights.h5",
903
+ "/layers/deepseek_vl_text_decoder_layer_3/attention/key/vars": "model_00000.weights.h5",
904
+ "/layers/deepseek_vl_text_decoder_layer_3/attention/output_proj/vars": "model_00000.weights.h5",
905
+ "/layers/deepseek_vl_text_decoder_layer_3/attention/query/vars": "model_00000.weights.h5",
906
+ "/layers/deepseek_vl_text_decoder_layer_3/attention/value/vars": "model_00000.weights.h5",
907
+ "/layers/deepseek_vl_text_decoder_layer_3/attention_norm/vars": "model_00000.weights.h5",
908
+ "/layers/deepseek_vl_text_decoder_layer_3/mlp/down/vars": "model_00000.weights.h5",
909
+ "/layers/deepseek_vl_text_decoder_layer_3/mlp/gate/vars": "model_00000.weights.h5",
910
+ "/layers/deepseek_vl_text_decoder_layer_3/mlp/up/vars": "model_00000.weights.h5",
911
+ "/layers/deepseek_vl_text_decoder_layer_3/mlp_norm/vars": "model_00000.weights.h5",
912
+ "/layers/deepseek_vl_text_decoder_layer_4/attention/key/vars": "model_00000.weights.h5",
913
+ "/layers/deepseek_vl_text_decoder_layer_4/attention/output_proj/vars": "model_00000.weights.h5",
914
+ "/layers/deepseek_vl_text_decoder_layer_4/attention/query/vars": "model_00000.weights.h5",
915
+ "/layers/deepseek_vl_text_decoder_layer_4/attention/value/vars": "model_00000.weights.h5",
916
+ "/layers/deepseek_vl_text_decoder_layer_4/attention_norm/vars": "model_00000.weights.h5",
917
+ "/layers/deepseek_vl_text_decoder_layer_4/mlp/down/vars": "model_00000.weights.h5",
918
+ "/layers/deepseek_vl_text_decoder_layer_4/mlp/gate/vars": "model_00000.weights.h5",
919
+ "/layers/deepseek_vl_text_decoder_layer_4/mlp/up/vars": "model_00000.weights.h5",
920
+ "/layers/deepseek_vl_text_decoder_layer_4/mlp_norm/vars": "model_00000.weights.h5",
921
+ "/layers/deepseek_vl_text_decoder_layer_5/attention/key/vars": "model_00000.weights.h5",
922
+ "/layers/deepseek_vl_text_decoder_layer_5/attention/output_proj/vars": "model_00000.weights.h5",
923
+ "/layers/deepseek_vl_text_decoder_layer_5/attention/query/vars": "model_00000.weights.h5",
924
+ "/layers/deepseek_vl_text_decoder_layer_5/attention/value/vars": "model_00000.weights.h5",
925
+ "/layers/deepseek_vl_text_decoder_layer_5/attention_norm/vars": "model_00000.weights.h5",
926
+ "/layers/deepseek_vl_text_decoder_layer_5/mlp/down/vars": "model_00000.weights.h5",
927
+ "/layers/deepseek_vl_text_decoder_layer_5/mlp/gate/vars": "model_00000.weights.h5",
928
+ "/layers/deepseek_vl_text_decoder_layer_5/mlp/up/vars": "model_00000.weights.h5",
929
+ "/layers/deepseek_vl_text_decoder_layer_5/mlp_norm/vars": "model_00000.weights.h5",
930
+ "/layers/deepseek_vl_text_decoder_layer_6/attention/key/vars": "model_00000.weights.h5",
931
+ "/layers/deepseek_vl_text_decoder_layer_6/attention/output_proj/vars": "model_00000.weights.h5",
932
+ "/layers/deepseek_vl_text_decoder_layer_6/attention/query/vars": "model_00000.weights.h5",
933
+ "/layers/deepseek_vl_text_decoder_layer_6/attention/value/vars": "model_00000.weights.h5",
934
+ "/layers/deepseek_vl_text_decoder_layer_6/attention_norm/vars": "model_00000.weights.h5",
935
+ "/layers/deepseek_vl_text_decoder_layer_6/mlp/down/vars": "model_00000.weights.h5",
936
+ "/layers/deepseek_vl_text_decoder_layer_6/mlp/gate/vars": "model_00000.weights.h5",
937
+ "/layers/deepseek_vl_text_decoder_layer_6/mlp/up/vars": "model_00000.weights.h5",
938
+ "/layers/deepseek_vl_text_decoder_layer_6/mlp_norm/vars": "model_00000.weights.h5",
939
+ "/layers/deepseek_vl_text_decoder_layer_7/attention/key/vars": "model_00000.weights.h5",
940
+ "/layers/deepseek_vl_text_decoder_layer_7/attention/output_proj/vars": "model_00000.weights.h5",
941
+ "/layers/deepseek_vl_text_decoder_layer_7/attention/query/vars": "model_00000.weights.h5",
942
+ "/layers/deepseek_vl_text_decoder_layer_7/attention/value/vars": "model_00000.weights.h5",
943
+ "/layers/deepseek_vl_text_decoder_layer_7/attention_norm/vars": "model_00000.weights.h5",
944
+ "/layers/deepseek_vl_text_decoder_layer_7/mlp/down/vars": "model_00000.weights.h5",
945
+ "/layers/deepseek_vl_text_decoder_layer_7/mlp/gate/vars": "model_00000.weights.h5",
946
+ "/layers/deepseek_vl_text_decoder_layer_7/mlp/up/vars": "model_00000.weights.h5",
947
+ "/layers/deepseek_vl_text_decoder_layer_7/mlp_norm/vars": "model_00000.weights.h5",
948
+ "/layers/deepseek_vl_text_decoder_layer_8/attention/key/vars": "model_00000.weights.h5",
949
+ "/layers/deepseek_vl_text_decoder_layer_8/attention/output_proj/vars": "model_00000.weights.h5",
950
+ "/layers/deepseek_vl_text_decoder_layer_8/attention/query/vars": "model_00000.weights.h5",
951
+ "/layers/deepseek_vl_text_decoder_layer_8/attention/value/vars": "model_00000.weights.h5",
952
+ "/layers/deepseek_vl_text_decoder_layer_8/attention_norm/vars": "model_00000.weights.h5",
953
+ "/layers/deepseek_vl_text_decoder_layer_8/mlp/down/vars": "model_00000.weights.h5",
954
+ "/layers/deepseek_vl_text_decoder_layer_8/mlp/gate/vars": "model_00000.weights.h5",
955
+ "/layers/deepseek_vl_text_decoder_layer_8/mlp/up/vars": "model_00000.weights.h5",
956
+ "/layers/deepseek_vl_text_decoder_layer_8/mlp_norm/vars": "model_00000.weights.h5",
957
+ "/layers/deepseek_vl_text_decoder_layer_9/attention/key/vars": "model_00000.weights.h5",
958
+ "/layers/deepseek_vl_text_decoder_layer_9/attention/output_proj/vars": "model_00001.weights.h5",
959
+ "/layers/deepseek_vl_text_decoder_layer_9/attention/query/vars": "model_00001.weights.h5",
960
+ "/layers/deepseek_vl_text_decoder_layer_9/attention/value/vars": "model_00001.weights.h5",
961
+ "/layers/deepseek_vl_text_decoder_layer_9/attention_norm/vars": "model_00001.weights.h5",
962
+ "/layers/deepseek_vl_text_decoder_layer_9/mlp/down/vars": "model_00001.weights.h5",
963
+ "/layers/deepseek_vl_text_decoder_layer_9/mlp/gate/vars": "model_00001.weights.h5",
964
+ "/layers/deepseek_vl_text_decoder_layer_9/mlp/up/vars": "model_00001.weights.h5",
965
+ "/layers/deepseek_vl_text_decoder_layer_9/mlp_norm/vars": "model_00001.weights.h5",
966
+ "/layers/deepseek_vl_text_decoder_layer_10/attention/key/vars": "model_00001.weights.h5",
967
+ "/layers/deepseek_vl_text_decoder_layer_10/attention/output_proj/vars": "model_00001.weights.h5",
968
+ "/layers/deepseek_vl_text_decoder_layer_10/attention/query/vars": "model_00001.weights.h5",
969
+ "/layers/deepseek_vl_text_decoder_layer_10/attention/value/vars": "model_00001.weights.h5",
970
+ "/layers/deepseek_vl_text_decoder_layer_10/attention_norm/vars": "model_00001.weights.h5",
971
+ "/layers/deepseek_vl_text_decoder_layer_10/mlp/down/vars": "model_00001.weights.h5",
972
+ "/layers/deepseek_vl_text_decoder_layer_10/mlp/gate/vars": "model_00001.weights.h5",
973
+ "/layers/deepseek_vl_text_decoder_layer_10/mlp/up/vars": "model_00001.weights.h5",
974
+ "/layers/deepseek_vl_text_decoder_layer_10/mlp_norm/vars": "model_00001.weights.h5",
975
+ "/layers/deepseek_vl_text_decoder_layer_11/attention/key/vars": "model_00001.weights.h5",
976
+ "/layers/deepseek_vl_text_decoder_layer_11/attention/output_proj/vars": "model_00001.weights.h5",
977
+ "/layers/deepseek_vl_text_decoder_layer_11/attention/query/vars": "model_00001.weights.h5",
978
+ "/layers/deepseek_vl_text_decoder_layer_11/attention/value/vars": "model_00001.weights.h5",
979
+ "/layers/deepseek_vl_text_decoder_layer_11/attention_norm/vars": "model_00001.weights.h5",
980
+ "/layers/deepseek_vl_text_decoder_layer_11/mlp/down/vars": "model_00001.weights.h5",
981
+ "/layers/deepseek_vl_text_decoder_layer_11/mlp/gate/vars": "model_00001.weights.h5",
982
+ "/layers/deepseek_vl_text_decoder_layer_11/mlp/up/vars": "model_00001.weights.h5",
983
+ "/layers/deepseek_vl_text_decoder_layer_11/mlp_norm/vars": "model_00001.weights.h5",
984
+ "/layers/deepseek_vl_text_decoder_layer_12/attention/key/vars": "model_00001.weights.h5",
985
+ "/layers/deepseek_vl_text_decoder_layer_12/attention/output_proj/vars": "model_00001.weights.h5",
986
+ "/layers/deepseek_vl_text_decoder_layer_12/attention/query/vars": "model_00001.weights.h5",
987
+ "/layers/deepseek_vl_text_decoder_layer_12/attention/value/vars": "model_00001.weights.h5",
988
+ "/layers/deepseek_vl_text_decoder_layer_12/attention_norm/vars": "model_00001.weights.h5",
989
+ "/layers/deepseek_vl_text_decoder_layer_12/mlp/down/vars": "model_00001.weights.h5",
990
+ "/layers/deepseek_vl_text_decoder_layer_12/mlp/gate/vars": "model_00001.weights.h5",
991
+ "/layers/deepseek_vl_text_decoder_layer_12/mlp/up/vars": "model_00001.weights.h5",
992
+ "/layers/deepseek_vl_text_decoder_layer_12/mlp_norm/vars": "model_00001.weights.h5",
993
+ "/layers/deepseek_vl_text_decoder_layer_13/attention/key/vars": "model_00001.weights.h5",
994
+ "/layers/deepseek_vl_text_decoder_layer_13/attention/output_proj/vars": "model_00001.weights.h5",
995
+ "/layers/deepseek_vl_text_decoder_layer_13/attention/query/vars": "model_00001.weights.h5",
996
+ "/layers/deepseek_vl_text_decoder_layer_13/attention/value/vars": "model_00001.weights.h5",
997
+ "/layers/deepseek_vl_text_decoder_layer_13/attention_norm/vars": "model_00001.weights.h5",
998
+ "/layers/deepseek_vl_text_decoder_layer_13/mlp/down/vars": "model_00001.weights.h5",
999
+ "/layers/deepseek_vl_text_decoder_layer_13/mlp/gate/vars": "model_00001.weights.h5",
1000
+ "/layers/deepseek_vl_text_decoder_layer_13/mlp/up/vars": "model_00001.weights.h5",
1001
+ "/layers/deepseek_vl_text_decoder_layer_13/mlp_norm/vars": "model_00001.weights.h5",
1002
+ "/layers/deepseek_vl_text_decoder_layer_14/attention/key/vars": "model_00001.weights.h5",
1003
+ "/layers/deepseek_vl_text_decoder_layer_14/attention/output_proj/vars": "model_00001.weights.h5",
1004
+ "/layers/deepseek_vl_text_decoder_layer_14/attention/query/vars": "model_00001.weights.h5",
1005
+ "/layers/deepseek_vl_text_decoder_layer_14/attention/value/vars": "model_00001.weights.h5",
1006
+ "/layers/deepseek_vl_text_decoder_layer_14/attention_norm/vars": "model_00001.weights.h5",
1007
+ "/layers/deepseek_vl_text_decoder_layer_14/mlp/down/vars": "model_00001.weights.h5",
1008
+ "/layers/deepseek_vl_text_decoder_layer_14/mlp/gate/vars": "model_00001.weights.h5",
1009
+ "/layers/deepseek_vl_text_decoder_layer_14/mlp/up/vars": "model_00001.weights.h5",
1010
+ "/layers/deepseek_vl_text_decoder_layer_14/mlp_norm/vars": "model_00001.weights.h5",
1011
+ "/layers/deepseek_vl_text_decoder_layer_15/attention/key/vars": "model_00001.weights.h5",
1012
+ "/layers/deepseek_vl_text_decoder_layer_15/attention/output_proj/vars": "model_00001.weights.h5",
1013
+ "/layers/deepseek_vl_text_decoder_layer_15/attention/query/vars": "model_00001.weights.h5",
1014
+ "/layers/deepseek_vl_text_decoder_layer_15/attention/value/vars": "model_00001.weights.h5",
1015
+ "/layers/deepseek_vl_text_decoder_layer_15/attention_norm/vars": "model_00001.weights.h5",
1016
+ "/layers/deepseek_vl_text_decoder_layer_15/mlp/down/vars": "model_00001.weights.h5",
1017
+ "/layers/deepseek_vl_text_decoder_layer_15/mlp/gate/vars": "model_00001.weights.h5",
1018
+ "/layers/deepseek_vl_text_decoder_layer_15/mlp/up/vars": "model_00001.weights.h5",
1019
+ "/layers/deepseek_vl_text_decoder_layer_15/mlp_norm/vars": "model_00001.weights.h5",
1020
+ "/layers/deepseek_vl_text_decoder_layer_16/attention/key/vars": "model_00001.weights.h5",
1021
+ "/layers/deepseek_vl_text_decoder_layer_16/attention/output_proj/vars": "model_00001.weights.h5",
1022
+ "/layers/deepseek_vl_text_decoder_layer_16/attention/query/vars": "model_00001.weights.h5",
1023
+ "/layers/deepseek_vl_text_decoder_layer_16/attention/value/vars": "model_00001.weights.h5",
1024
+ "/layers/deepseek_vl_text_decoder_layer_16/attention_norm/vars": "model_00001.weights.h5",
1025
+ "/layers/deepseek_vl_text_decoder_layer_16/mlp/down/vars": "model_00001.weights.h5",
1026
+ "/layers/deepseek_vl_text_decoder_layer_16/mlp/gate/vars": "model_00001.weights.h5",
1027
+ "/layers/deepseek_vl_text_decoder_layer_16/mlp/up/vars": "model_00001.weights.h5",
1028
+ "/layers/deepseek_vl_text_decoder_layer_16/mlp_norm/vars": "model_00001.weights.h5",
1029
+ "/layers/deepseek_vl_text_decoder_layer_17/attention/key/vars": "model_00001.weights.h5",
1030
+ "/layers/deepseek_vl_text_decoder_layer_17/attention/output_proj/vars": "model_00001.weights.h5",
1031
+ "/layers/deepseek_vl_text_decoder_layer_17/attention/query/vars": "model_00001.weights.h5",
1032
+ "/layers/deepseek_vl_text_decoder_layer_17/attention/value/vars": "model_00001.weights.h5",
1033
+ "/layers/deepseek_vl_text_decoder_layer_17/attention_norm/vars": "model_00001.weights.h5",
1034
+ "/layers/deepseek_vl_text_decoder_layer_17/mlp/down/vars": "model_00001.weights.h5",
1035
+ "/layers/deepseek_vl_text_decoder_layer_17/mlp/gate/vars": "model_00001.weights.h5",
1036
+ "/layers/deepseek_vl_text_decoder_layer_17/mlp/up/vars": "model_00001.weights.h5",
1037
+ "/layers/deepseek_vl_text_decoder_layer_17/mlp_norm/vars": "model_00001.weights.h5",
1038
+ "/layers/deepseek_vl_text_decoder_layer_18/attention/key/vars": "model_00001.weights.h5",
1039
+ "/layers/deepseek_vl_text_decoder_layer_18/attention/output_proj/vars": "model_00001.weights.h5",
1040
+ "/layers/deepseek_vl_text_decoder_layer_18/attention/query/vars": "model_00001.weights.h5",
1041
+ "/layers/deepseek_vl_text_decoder_layer_18/attention/value/vars": "model_00001.weights.h5",
1042
+ "/layers/deepseek_vl_text_decoder_layer_18/attention_norm/vars": "model_00001.weights.h5",
1043
+ "/layers/deepseek_vl_text_decoder_layer_18/mlp/down/vars": "model_00001.weights.h5",
1044
+ "/layers/deepseek_vl_text_decoder_layer_18/mlp/gate/vars": "model_00001.weights.h5",
1045
+ "/layers/deepseek_vl_text_decoder_layer_18/mlp/up/vars": "model_00001.weights.h5",
1046
+ "/layers/deepseek_vl_text_decoder_layer_18/mlp_norm/vars": "model_00001.weights.h5",
1047
+ "/layers/deepseek_vl_text_decoder_layer_19/attention/key/vars": "model_00001.weights.h5",
1048
+ "/layers/deepseek_vl_text_decoder_layer_19/attention/output_proj/vars": "model_00001.weights.h5",
1049
+ "/layers/deepseek_vl_text_decoder_layer_19/attention/query/vars": "model_00001.weights.h5",
1050
+ "/layers/deepseek_vl_text_decoder_layer_19/attention/value/vars": "model_00001.weights.h5",
1051
+ "/layers/deepseek_vl_text_decoder_layer_19/attention_norm/vars": "model_00001.weights.h5",
1052
+ "/layers/deepseek_vl_text_decoder_layer_19/mlp/down/vars": "model_00001.weights.h5",
1053
+ "/layers/deepseek_vl_text_decoder_layer_19/mlp/gate/vars": "model_00001.weights.h5",
1054
+ "/layers/deepseek_vl_text_decoder_layer_19/mlp/up/vars": "model_00001.weights.h5",
1055
+ "/layers/deepseek_vl_text_decoder_layer_19/mlp_norm/vars": "model_00001.weights.h5",
1056
+ "/layers/deepseek_vl_text_decoder_layer_20/attention/key/vars": "model_00001.weights.h5",
1057
+ "/layers/deepseek_vl_text_decoder_layer_20/attention/output_proj/vars": "model_00001.weights.h5",
1058
+ "/layers/deepseek_vl_text_decoder_layer_20/attention/query/vars": "model_00001.weights.h5",
1059
+ "/layers/deepseek_vl_text_decoder_layer_20/attention/value/vars": "model_00001.weights.h5",
1060
+ "/layers/deepseek_vl_text_decoder_layer_20/attention_norm/vars": "model_00001.weights.h5",
1061
+ "/layers/deepseek_vl_text_decoder_layer_20/mlp/down/vars": "model_00001.weights.h5",
1062
+ "/layers/deepseek_vl_text_decoder_layer_20/mlp/gate/vars": "model_00001.weights.h5",
1063
+ "/layers/deepseek_vl_text_decoder_layer_20/mlp/up/vars": "model_00001.weights.h5",
1064
+ "/layers/deepseek_vl_text_decoder_layer_20/mlp_norm/vars": "model_00001.weights.h5",
1065
+ "/layers/deepseek_vl_text_decoder_layer_21/attention/key/vars": "model_00001.weights.h5",
1066
+ "/layers/deepseek_vl_text_decoder_layer_21/attention/output_proj/vars": "model_00001.weights.h5",
1067
+ "/layers/deepseek_vl_text_decoder_layer_21/attention/query/vars": "model_00001.weights.h5",
1068
+ "/layers/deepseek_vl_text_decoder_layer_21/attention/value/vars": "model_00001.weights.h5",
1069
+ "/layers/deepseek_vl_text_decoder_layer_21/attention_norm/vars": "model_00001.weights.h5",
1070
+ "/layers/deepseek_vl_text_decoder_layer_21/mlp/down/vars": "model_00001.weights.h5",
1071
+ "/layers/deepseek_vl_text_decoder_layer_21/mlp/gate/vars": "model_00001.weights.h5",
1072
+ "/layers/deepseek_vl_text_decoder_layer_21/mlp/up/vars": "model_00001.weights.h5",
1073
+ "/layers/deepseek_vl_text_decoder_layer_21/mlp_norm/vars": "model_00001.weights.h5",
1074
+ "/layers/deepseek_vl_text_decoder_layer_22/attention/key/vars": "model_00001.weights.h5",
1075
+ "/layers/deepseek_vl_text_decoder_layer_22/attention/output_proj/vars": "model_00001.weights.h5",
1076
+ "/layers/deepseek_vl_text_decoder_layer_22/attention/query/vars": "model_00001.weights.h5",
1077
+ "/layers/deepseek_vl_text_decoder_layer_22/attention/value/vars": "model_00001.weights.h5",
1078
+ "/layers/deepseek_vl_text_decoder_layer_22/attention_norm/vars": "model_00001.weights.h5",
1079
+ "/layers/deepseek_vl_text_decoder_layer_22/mlp/down/vars": "model_00002.weights.h5",
1080
+ "/layers/deepseek_vl_text_decoder_layer_22/mlp/gate/vars": "model_00002.weights.h5",
1081
+ "/layers/deepseek_vl_text_decoder_layer_22/mlp/up/vars": "model_00002.weights.h5",
1082
+ "/layers/deepseek_vl_text_decoder_layer_22/mlp_norm/vars": "model_00002.weights.h5",
1083
+ "/layers/deepseek_vl_text_decoder_layer_23/attention/key/vars": "model_00002.weights.h5",
1084
+ "/layers/deepseek_vl_text_decoder_layer_23/attention/output_proj/vars": "model_00002.weights.h5",
1085
+ "/layers/deepseek_vl_text_decoder_layer_23/attention/query/vars": "model_00002.weights.h5",
1086
+ "/layers/deepseek_vl_text_decoder_layer_23/attention/value/vars": "model_00002.weights.h5",
1087
+ "/layers/deepseek_vl_text_decoder_layer_23/attention_norm/vars": "model_00002.weights.h5",
1088
+ "/layers/deepseek_vl_text_decoder_layer_23/mlp/down/vars": "model_00002.weights.h5",
1089
+ "/layers/deepseek_vl_text_decoder_layer_23/mlp/gate/vars": "model_00002.weights.h5",
1090
+ "/layers/deepseek_vl_text_decoder_layer_23/mlp/up/vars": "model_00002.weights.h5",
1091
+ "/layers/deepseek_vl_text_decoder_layer_23/mlp_norm/vars": "model_00002.weights.h5",
1092
+ "/layers/deepseek_vl_text_decoder_layer_24/attention/key/vars": "model_00002.weights.h5",
1093
+ "/layers/deepseek_vl_text_decoder_layer_24/attention/output_proj/vars": "model_00002.weights.h5",
1094
+ "/layers/deepseek_vl_text_decoder_layer_24/attention/query/vars": "model_00002.weights.h5",
1095
+ "/layers/deepseek_vl_text_decoder_layer_24/attention/value/vars": "model_00002.weights.h5",
1096
+ "/layers/deepseek_vl_text_decoder_layer_24/attention_norm/vars": "model_00002.weights.h5",
1097
+ "/layers/deepseek_vl_text_decoder_layer_24/mlp/down/vars": "model_00002.weights.h5",
1098
+ "/layers/deepseek_vl_text_decoder_layer_24/mlp/gate/vars": "model_00002.weights.h5",
1099
+ "/layers/deepseek_vl_text_decoder_layer_24/mlp/up/vars": "model_00002.weights.h5",
1100
+ "/layers/deepseek_vl_text_decoder_layer_24/mlp_norm/vars": "model_00002.weights.h5",
1101
+ "/layers/deepseek_vl_text_decoder_layer_25/attention/key/vars": "model_00002.weights.h5",
1102
+ "/layers/deepseek_vl_text_decoder_layer_25/attention/output_proj/vars": "model_00002.weights.h5",
1103
+ "/layers/deepseek_vl_text_decoder_layer_25/attention/query/vars": "model_00002.weights.h5",
1104
+ "/layers/deepseek_vl_text_decoder_layer_25/attention/value/vars": "model_00002.weights.h5",
1105
+ "/layers/deepseek_vl_text_decoder_layer_25/attention_norm/vars": "model_00002.weights.h5",
1106
+ "/layers/deepseek_vl_text_decoder_layer_25/mlp/down/vars": "model_00002.weights.h5",
1107
+ "/layers/deepseek_vl_text_decoder_layer_25/mlp/gate/vars": "model_00002.weights.h5",
1108
+ "/layers/deepseek_vl_text_decoder_layer_25/mlp/up/vars": "model_00002.weights.h5",
1109
+ "/layers/deepseek_vl_text_decoder_layer_25/mlp_norm/vars": "model_00002.weights.h5",
1110
+ "/layers/deepseek_vl_text_decoder_layer_26/attention/key/vars": "model_00002.weights.h5",
1111
+ "/layers/deepseek_vl_text_decoder_layer_26/attention/output_proj/vars": "model_00002.weights.h5",
1112
+ "/layers/deepseek_vl_text_decoder_layer_26/attention/query/vars": "model_00002.weights.h5",
1113
+ "/layers/deepseek_vl_text_decoder_layer_26/attention/value/vars": "model_00002.weights.h5",
1114
+ "/layers/deepseek_vl_text_decoder_layer_26/attention_norm/vars": "model_00002.weights.h5",
1115
+ "/layers/deepseek_vl_text_decoder_layer_26/mlp/down/vars": "model_00002.weights.h5",
1116
+ "/layers/deepseek_vl_text_decoder_layer_26/mlp/gate/vars": "model_00002.weights.h5",
1117
+ "/layers/deepseek_vl_text_decoder_layer_26/mlp/up/vars": "model_00002.weights.h5",
1118
+ "/layers/deepseek_vl_text_decoder_layer_26/mlp_norm/vars": "model_00002.weights.h5",
1119
+ "/layers/deepseek_vl_text_decoder_layer_27/attention/key/vars": "model_00002.weights.h5",
1120
+ "/layers/deepseek_vl_text_decoder_layer_27/attention/output_proj/vars": "model_00002.weights.h5",
1121
+ "/layers/deepseek_vl_text_decoder_layer_27/attention/query/vars": "model_00002.weights.h5",
1122
+ "/layers/deepseek_vl_text_decoder_layer_27/attention/value/vars": "model_00002.weights.h5",
1123
+ "/layers/deepseek_vl_text_decoder_layer_27/attention_norm/vars": "model_00002.weights.h5",
1124
+ "/layers/deepseek_vl_text_decoder_layer_27/mlp/down/vars": "model_00002.weights.h5",
1125
+ "/layers/deepseek_vl_text_decoder_layer_27/mlp/gate/vars": "model_00002.weights.h5",
1126
+ "/layers/deepseek_vl_text_decoder_layer_27/mlp/up/vars": "model_00002.weights.h5",
1127
+ "/layers/deepseek_vl_text_decoder_layer_27/mlp_norm/vars": "model_00002.weights.h5",
1128
+ "/layers/deepseek_vl_text_decoder_layer_28/attention/key/vars": "model_00002.weights.h5",
1129
+ "/layers/deepseek_vl_text_decoder_layer_28/attention/output_proj/vars": "model_00002.weights.h5",
1130
+ "/layers/deepseek_vl_text_decoder_layer_28/attention/query/vars": "model_00002.weights.h5",
1131
+ "/layers/deepseek_vl_text_decoder_layer_28/attention/value/vars": "model_00002.weights.h5",
1132
+ "/layers/deepseek_vl_text_decoder_layer_28/attention_norm/vars": "model_00002.weights.h5",
1133
+ "/layers/deepseek_vl_text_decoder_layer_28/mlp/down/vars": "model_00002.weights.h5",
1134
+ "/layers/deepseek_vl_text_decoder_layer_28/mlp/gate/vars": "model_00002.weights.h5",
1135
+ "/layers/deepseek_vl_text_decoder_layer_28/mlp/up/vars": "model_00002.weights.h5",
1136
+ "/layers/deepseek_vl_text_decoder_layer_28/mlp_norm/vars": "model_00002.weights.h5",
1137
+ "/layers/deepseek_vl_text_decoder_layer_29/attention/key/vars": "model_00002.weights.h5",
1138
+ "/layers/deepseek_vl_text_decoder_layer_29/attention/output_proj/vars": "model_00002.weights.h5",
1139
+ "/layers/deepseek_vl_text_decoder_layer_29/attention/query/vars": "model_00002.weights.h5",
1140
+ "/layers/deepseek_vl_text_decoder_layer_29/attention/value/vars": "model_00002.weights.h5",
1141
+ "/layers/deepseek_vl_text_decoder_layer_29/attention_norm/vars": "model_00002.weights.h5",
1142
+ "/layers/deepseek_vl_text_decoder_layer_29/mlp/down/vars": "model_00002.weights.h5",
1143
+ "/layers/deepseek_vl_text_decoder_layer_29/mlp/gate/vars": "model_00002.weights.h5",
1144
+ "/layers/deepseek_vl_text_decoder_layer_29/mlp/up/vars": "model_00002.weights.h5",
1145
+ "/layers/deepseek_vl_text_decoder_layer_29/mlp_norm/vars": "model_00002.weights.h5",
1146
+ "/layers/deepseek_vl_text_rms_norm/vars": "model_00002.weights.h5"
1147
+ }
1148
+ }
model_00000.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:238756cfb6897451a4c6f33b0250b23570ab787a71fa2d71c8de7c04e13de067
3
+ size 5358603448
model_00001.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98064d33bf4bb65137eea3547e6cddc1dc6d84e757063176bba5368fd96fb30d
3
+ size 5363021928
model_00002.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e4ec91b2c50012eae0a9c3a9a3d4f54f9f1849b1e0cb33aca3cb5da59b58d5f
3
+ size 3104129152
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff