Add DeepSeek-OCR-2 fp16 ONNX export
Browse files- .gitattributes +4 -0
- deepseek-ocr2-fp16/config.json +45 -0
- deepseek-ocr2-fp16/decoder.onnx +3 -0
- deepseek-ocr2-fp16/decoder.onnx.data +3 -0
- deepseek-ocr2-fp16/tokenizer.json +0 -0
- deepseek-ocr2-fp16/vision_encoder.onnx +3 -0
- deepseek-ocr2-fp16/vision_encoder.onnx.data +3 -0
- deepseek-ocr2-fp16/vision_sam_1024.onnx +3 -0
- deepseek-ocr2-fp16/vision_sam_1024.onnx.data +3 -0
- deepseek-ocr2-fp16/vision_sam_768.onnx +3 -0
- deepseek-ocr2-fp16/vision_sam_768.onnx.data +3 -0
.gitattributes
CHANGED
|
@@ -35,3 +35,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
opus-mt-es-en/source.spm filter=lfs diff=lfs merge=lfs -text
|
| 37 |
opus-mt-es-en/target.spm filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
opus-mt-es-en/source.spm filter=lfs diff=lfs merge=lfs -text
|
| 37 |
opus-mt-es-en/target.spm filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
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
|
deepseek-ocr2-fp16/config.json
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "deepseek-ai/DeepSeek-OCR-2",
|
| 3 |
+
"dtype": "fp16",
|
| 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-fp16/decoder.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:59954393d04669340e9f44c31921829af7f648235fe539d402208e4f0141511b
|
| 3 |
+
size 219619
|
deepseek-ocr2-fp16/decoder.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:856ca8ad735db70a0e1cef8fd5a4ef8f2df96157917ac775596b1b7e874261a5
|
| 3 |
+
size 5871272960
|
deepseek-ocr2-fp16/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
deepseek-ocr2-fp16/vision_encoder.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc37b858c1e7f018b8321982a37e192dfdeee472c2dbe824bb193ae6df75a391
|
| 3 |
+
size 910211
|
deepseek-ocr2-fp16/vision_encoder.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b2422d295d06abd7554d4492ee2509452c9ac03ad7ca02012e896febf408465d
|
| 3 |
+
size 718797056
|
deepseek-ocr2-fp16/vision_sam_1024.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2c93164e2f6a2efb225079eb9f2a460ce9b83669e6b22fbf1b49d822688518d2
|
| 3 |
+
size 599874
|
deepseek-ocr2-fp16/vision_sam_1024.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e49d1c2dc8e3518ba944a925fb2778d67058ed1e937f3555266923d4717a46c
|
| 3 |
+
size 189956608
|
deepseek-ocr2-fp16/vision_sam_768.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ef7de0d268fcc6881fe5db544cd85f3212f449b1162f1c8a0058c89e72ab037b
|
| 3 |
+
size 606685
|
deepseek-ocr2-fp16/vision_sam_768.onnx.data
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1033eb59e3dab7ce84e758e90dbb7fef4b39b2ff74efa8868bf74398ea810f9f
|
| 3 |
+
size 187334144
|