Add DeepSeek-OCR-2 fp32 ONNX export
Browse files- .gitattributes +4 -0
- deepseek-ocr2-fp32/config.json +45 -0
- deepseek-ocr2-fp32/decoder.onnx +3 -0
- deepseek-ocr2-fp32/decoder.onnx.data +3 -0
- deepseek-ocr2-fp32/tokenizer.json +0 -0
- deepseek-ocr2-fp32/vision_encoder.onnx +3 -0
- deepseek-ocr2-fp32/vision_encoder.onnx.data +3 -0
- deepseek-ocr2-fp32/vision_sam_1024.onnx +3 -0
- deepseek-ocr2-fp32/vision_sam_1024.onnx.data +3 -0
- deepseek-ocr2-fp32/vision_sam_768.onnx +3 -0
- deepseek-ocr2-fp32/vision_sam_768.onnx.data +3 -0
.gitattributes
CHANGED
|
@@ -39,3 +39,7 @@ deepseek-ocr2-fp16/decoder.onnx.data filter=lfs diff=lfs merge=lfs -text
|
|
| 39 |
deepseek-ocr2-fp16/vision_encoder.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 40 |
deepseek-ocr2-fp16/vision_sam_1024.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 41 |
deepseek-ocr2-fp16/vision_sam_768.onnx.data filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
deepseek-ocr2-fp16/vision_encoder.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 40 |
deepseek-ocr2-fp16/vision_sam_1024.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 41 |
deepseek-ocr2-fp16/vision_sam_768.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 42 |
+
deepseek-ocr2-fp32/decoder.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 43 |
+
deepseek-ocr2-fp32/vision_encoder.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 44 |
+
deepseek-ocr2-fp32/vision_sam_1024.onnx.data filter=lfs diff=lfs merge=lfs -text
|
| 45 |
+
deepseek-ocr2-fp32/vision_sam_768.onnx.data filter=lfs diff=lfs merge=lfs -text
|
deepseek-ocr2-fp32/config.json
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "deepseek-ai/DeepSeek-OCR-2",
|
| 3 |
+
"dtype": "fp32",
|
| 4 |
+
"num_layers": 12,
|
| 5 |
+
"num_heads": 10,
|
| 6 |
+
"head_dim": 128,
|
| 7 |
+
"hidden_size": 1280,
|
| 8 |
+
"vocab_size": 129280,
|
| 9 |
+
"bos_token_id": 0,
|
| 10 |
+
"eos_token_id": 1,
|
| 11 |
+
"image_token_id": 128815,
|
| 12 |
+
"base_size": 1024,
|
| 13 |
+
"crop_size": 768,
|
| 14 |
+
"base_tokens": 256,
|
| 15 |
+
"crop_tokens": 144,
|
| 16 |
+
"min_crops": 2,
|
| 17 |
+
"max_crops": 6,
|
| 18 |
+
"prompts": {
|
| 19 |
+
"markdown": {
|
| 20 |
+
"text": "<image>\n<|grounding|>Convert the document to markdown.",
|
| 21 |
+
"prefix_ids": [],
|
| 22 |
+
"suffix_ids": [
|
| 23 |
+
201,
|
| 24 |
+
128820,
|
| 25 |
+
21842,
|
| 26 |
+
270,
|
| 27 |
+
4940,
|
| 28 |
+
304,
|
| 29 |
+
2121,
|
| 30 |
+
7919,
|
| 31 |
+
16
|
| 32 |
+
]
|
| 33 |
+
},
|
| 34 |
+
"ocr": {
|
| 35 |
+
"text": "<image>\nFree OCR.",
|
| 36 |
+
"prefix_ids": [],
|
| 37 |
+
"suffix_ids": [
|
| 38 |
+
201,
|
| 39 |
+
21431,
|
| 40 |
+
126041,
|
| 41 |
+
16
|
| 42 |
+
]
|
| 43 |
+
}
|
| 44 |
+
}
|
| 45 |
+
}
|
deepseek-ocr2-fp32/decoder.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12ab1aef9dc58b5eae260432a5253ae911cd75740b225a0a4debb1026e15bc08
|
| 3 |
+
size 219196
|
deepseek-ocr2-fp32/decoder.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2ef090583fff0a185a172ae7d0d7342e6e8227129a57af15cd0d736547aa3ec
|
| 3 |
+
size 11738941440
|
deepseek-ocr2-fp32/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
deepseek-ocr2-fp32/vision_encoder.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb7af4b2841e4572b279c671673e73e8729325c0474c184e1e979c41e9faebad
|
| 3 |
+
size 920821
|
deepseek-ocr2-fp32/vision_encoder.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:39f383caea83b7da6a7e438a4c0fc5a3af2e6b986480361e953879cc10e64c44
|
| 3 |
+
size 1437594112
|
deepseek-ocr2-fp32/vision_sam_1024.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0e939b213cfcf8372b271d6a14a54effd1b7886ffa2217eb31af80f76a765f4
|
| 3 |
+
size 596897
|
deepseek-ocr2-fp32/vision_sam_1024.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0c540ba1296e7463170f0b4b0d6af22ad566254176bea145aea9c85e8eda283c
|
| 3 |
+
size 379917312
|
deepseek-ocr2-fp32/vision_sam_768.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f8b99a22ce5aed9b852fff1dafda96ece7f5e05f70fb3bd67dab37addfbd46ae
|
| 3 |
+
size 603589
|
deepseek-ocr2-fp32/vision_sam_768.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95a07aec108afb8e3192d5a66f59625b89ae1ca471bfc072ac4377aad65076d4
|
| 3 |
+
size 374412288
|