2026-09-26: PP-DocLayoutV3_fp32_batchable.onnx, PP-DocLayoutV3_fp32_batchable_fp16.onnx, PP-OCRv6_medium_det.onnx, PP-OCRv6_small_det.onnx, PP-OCRv6_tiny_det.onnx, PP-OCRv6_tiny_rec_ctc.onnx, PP-OCRv6_tiny_rec_ctc_fp16.onnx, PP-OCRv6_medium_rec_ctc.onnx, PP-OCRv6_medium_rec_ctc_fp16.onnx, PP-OCRv6_small_rec_ctc.onnx, PP-OCRv6_small_rec_ctc_fp16.onnx, PP-FormulaNet_plus-M_encoder.onnx, PP-FormulaNet_plus-M_prep.onnx, PP-FormulaNet_plus-M_decoder_step_kv256.onnx, PP-FormulaNet_plus-M_decoder_step_kv512.onnx, PP-FormulaNet_plus-M_decoder_step_kv1536.onnx, PP-FormulaNet_plus-M_kv_cast.onnx, PP-FormulaNet_plus-M_kv_scatter.onnx, PP-FormulaNet_plus-M_kv_load.onnx, PP-FormulaNet_plus-M_kv_take.onnx, PP-FormulaNet_plus-M_kv_put.onnx, PP-FormulaNet_plus-M_tokenizer.json, PP-FormulaNet_plus-S_encoder.onnx, PP-FormulaNet_plus-S_prep.onnx, PP-FormulaNet_plus-S_decoder_step_kv256.onnx, PP-FormulaNet_plus-S_decoder_step_kv512.onnx, PP-FormulaNet_plus-S_decoder_step_kv1536.onnx, PP-FormulaNet_plus-S_kv_cast.onnx, PP-FormulaNet_plus-S_kv_scatter.onnx, PP-FormulaNet_plus-S_kv_load.onnx, PP-FormulaNet_plus-S_kv_take.onnx, PP-FormulaNet_plus-S_kv_put.onnx, PP-FormulaNet_plus-S_tokenizer.json, PP-LCNet_x1_0_table_cls.onnx, RapidTable_unet_uint8.onnx, SLANet_plus_encoder.onnx, SLANet_plus_decoder.bin

MANIFEST.txt

@@ -11,11 +11,25 @@ ed513aba775144c855d5864cc52137a767f186124e23db5945dd41ca518f69ac  PP-OCRv6_small
 30490f7b582c233248ba39b21bbd3d99cd0b89c0c444409ce8c5f5aee178f866  PP-OCRv6_small_rec_ctc_fp16.onnx
 e15dc1f333c680ba2b9eef5a8be713391ad9b587c074ff5120254063ea7a1244  PP-FormulaNet_plus-M_encoder.onnx
 6fd37a4dc090b28c1a2a62516ec863786f8033dca14638400c0f7e3ce71d061f  PP-FormulaNet_plus-M_prep.onnx
-28b622239781155ab646917ac5cc0b15d32f24cebb59a7cd7a996bd6c7fdf789  PP-FormulaNet_plus-M_decoder_step.onnx
+6e9f396867cbc11b91a787d04c60590aa8c109c54141b691b71f6e8c56a5f2c2  PP-FormulaNet_plus-M_decoder_step_kv256.onnx
+d0208c24ada21e29b5956917e71bc48b5078a91712ee427852a7fafb7d079ccf  PP-FormulaNet_plus-M_decoder_step_kv512.onnx
+3571bf2937716776e03b87eb6866bffc6a568b70a3ec60a1f47d6a502634f4bd  PP-FormulaNet_plus-M_decoder_step_kv1536.onnx
+080f992a9a549847a5b8a45e26aa3e588e112c6c8466f640fa1cf3f114d44cb1  PP-FormulaNet_plus-M_kv_cast.onnx
+f26689afb17dcb3657c814589a9325d42528031e124160eeff836476c325fee2  PP-FormulaNet_plus-M_kv_scatter.onnx
+c2b0d008450375817c88b4cad510f4ce2eebd8d8d699b9a56b8f0fc22da05106  PP-FormulaNet_plus-M_kv_load.onnx
+53dfbabda07767c11dea5fca4451645f23f1c67cbe430c8dd8bdf2c2513ca4d5  PP-FormulaNet_plus-M_kv_take.onnx
+72bba68ba7e3234b57f6344731aaab9a2a7f9e85daec779784c82f079eda061c  PP-FormulaNet_plus-M_kv_put.onnx
 2811d82701ec97c192fa256aa2b4516929373870ae660326cc5b1dc879b95ff2  PP-FormulaNet_plus-M_tokenizer.json
 a021388d62aa43953b1dc7a3e0ea919b076aa8fdb98d8491830c6dbf2e5e4706  PP-FormulaNet_plus-S_encoder.onnx
 c18cf4eb94c3d16cf88faad285333a5b4b30b6e1718c2b92b4564baf984fbd7e  PP-FormulaNet_plus-S_prep.onnx
-e424f4400c921496eefc63a313b61c2167497f4a3c758cfc8dc40686ef6f5036  PP-FormulaNet_plus-S_decoder_step.onnx
+b9a6346ee885816f0892178c0504ce37880be0db195044291ef71e8fc29f6409  PP-FormulaNet_plus-S_decoder_step_kv256.onnx
+a789ed6bf8542fa071f8f4a1e73ec99e21836b7cfcb6a29916e7966cd4b91d0b  PP-FormulaNet_plus-S_decoder_step_kv512.onnx
+65b2ab562767a8994b016eb12fc490ca9bfa9297d1398247481bb325fef61f8e  PP-FormulaNet_plus-S_decoder_step_kv1536.onnx
+681ea9414b5fd37c69868f4b431b49134bfe1dd7051c51ecc704206ca0c5b654  PP-FormulaNet_plus-S_kv_cast.onnx
+79d93a4836dc6626ac1f7c1f2ba4f9cb0ab0eb777bd3da1752421d8696493a14  PP-FormulaNet_plus-S_kv_scatter.onnx
+628633c4ebed693ebfa1035ed153bd2ed5d7f3cc14f481b34659c02e12dec38e  PP-FormulaNet_plus-S_kv_load.onnx
+f45938e4ab5573911f70e247e6209f48f5a78446b941ba414a1f64e23aaf6255  PP-FormulaNet_plus-S_kv_take.onnx
+f02d86fb8943f792a96a0253eb7a7bac90c721522ffbd161c244f7bc6fc2a320  PP-FormulaNet_plus-S_kv_put.onnx
 2811d82701ec97c192fa256aa2b4516929373870ae660326cc5b1dc879b95ff2  PP-FormulaNet_plus-S_tokenizer.json
 f8e4cb1b58a29bebd36852edcb238b53e0acefa75116a8a4db443f13dbc72b0b  PP-LCNet_x1_0_table_cls.onnx
 69737cded76e1106e85585547995cc952c6e98a978402cf916f5de9582e773ad  RapidTable_unet_uint8.onnx

PP-FormulaNet_plus-M_decoder_step_kv1536.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3571bf2937716776e03b87eb6866bffc6a568b70a3ec60a1f47d6a502634f4bd
+size 149318039

PP-FormulaNet_plus-M_decoder_step_kv256.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e9f396867cbc11b91a787d04c60590aa8c109c54141b691b71f6e8c56a5f2c2
+size 149307799

PP-FormulaNet_plus-M_decoder_step_kv512.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0208c24ada21e29b5956917e71bc48b5078a91712ee427852a7fafb7d079ccf
+size 149309847

PP-FormulaNet_plus-M_kv_cast.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:080f992a9a549847a5b8a45e26aa3e588e112c6c8466f640fa1cf3f114d44cb1
+size 226

PP-FormulaNet_plus-M_kv_load.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2b0d008450375817c88b4cad510f4ce2eebd8d8d699b9a56b8f0fc22da05106
+size 724

PP-FormulaNet_plus-M_kv_put.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bba68ba7e3234b57f6344731aaab9a2a7f9e85daec779784c82f079eda061c
+size 1094

PP-FormulaNet_plus-M_kv_scatter.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f26689afb17dcb3657c814589a9325d42528031e124160eeff836476c325fee2
+size 659

PP-FormulaNet_plus-M_kv_take.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53dfbabda07767c11dea5fca4451645f23f1c67cbe430c8dd8bdf2c2513ca4d5
+size 277

PP-FormulaNet_plus-S_decoder_step_kv1536.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65b2ab562767a8994b016eb12fc490ca9bfa9297d1398247481bb325fef61f8e
+size 85937685

PP-FormulaNet_plus-S_decoder_step_kv256.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9a6346ee885816f0892178c0504ce37880be0db195044291ef71e8fc29f6409
+size 85927445

PP-FormulaNet_plus-S_decoder_step_kv512.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a789ed6bf8542fa071f8f4a1e73ec99e21836b7cfcb6a29916e7966cd4b91d0b
+size 85929493

PP-FormulaNet_plus-S_kv_cast.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:681ea9414b5fd37c69868f4b431b49134bfe1dd7051c51ecc704206ca0c5b654
+size 226

PP-FormulaNet_plus-S_kv_load.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:628633c4ebed693ebfa1035ed153bd2ed5d7f3cc14f481b34659c02e12dec38e
+size 724

PP-FormulaNet_plus-S_kv_put.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f02d86fb8943f792a96a0253eb7a7bac90c721522ffbd161c244f7bc6fc2a320
+size 1094

PP-FormulaNet_plus-S_kv_scatter.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79d93a4836dc6626ac1f7c1f2ba4f9cb0ab0eb777bd3da1752421d8696493a14
+size 675

PP-FormulaNet_plus-S_kv_take.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f45938e4ab5573911f70e247e6209f48f5a78446b941ba414a1f64e23aaf6255
+size 277

README.md

@@ -22,7 +22,7 @@ ONNX exports of PaddlePaddle document models, one flat file per model.
 | `SLANet_plus_encoder.onnx`, `SLANet_plus_decoder.bin` | table structure | [`PaddlePaddle/SLANet_plus_onnx`](https://huggingface.co/PaddlePaddle/SLANet_plus_onnx): the encoder graph, and the GRU decoder's weights as raw float32 |
 | `PP-LCNet_x1_0_table_cls.onnx` | table classification (wired / wireless) | [`PaddlePaddle/PP-LCNet_x1_0_table_cls_onnx`](https://huggingface.co/PaddlePaddle/PP-LCNet_x1_0_table_cls_onnx), unchanged; labels `[wired_table, wireless_table]` |
 | `RapidTable_unet_uint8.onnx` | wired-table line segmentation | RapidAI [TableStructureRec](https://github.com/RapidAI/TableStructureRec)'s `unet.onnx` (Apache-2.0; byte-identical in [`opendatalab/PDF-Extract-Kit-1.0`](https://huggingface.co/opendatalab/PDF-Extract-Kit-1.0) `models/TabRec/UnetStructure/`), with a UINT8 RGB `[1,H,W,3]` input normalised in the graph and a UINT8 `[H,W]` class map output (`unet_io.py`) |
-| `PP-FormulaNet_plus-{M,S}_{encoder,prep,decoder_step}.onnx`, `_tokenizer.json` | formula recognition | [`PaddlePaddle/PP-FormulaNet_plus-M`](https://huggingface.co/PaddlePaddle/PP-FormulaNet_plus-M) / [`-S`](https://huggingface.co/PaddlePaddle/PP-FormulaNet_plus-S), split into encoder, cross-attention precompute and one greedy decoder step over a static KV cache |
+| `PP-FormulaNet_plus-{M,S}_{encoder,prep,decoder_step_kv{256,512,1536},kv_cast,kv_scatter,kv_load,kv_take,kv_put}.onnx`, `_tokenizer.json` | formula recognition | [`PaddlePaddle/PP-FormulaNet_plus-M`](https://huggingface.co/PaddlePaddle/PP-FormulaNet_plus-M) / [`-S`](https://huggingface.co/PaddlePaddle/PP-FormulaNet_plus-S), split into encoder, cross-attention precompute and one greedy decoder step over a static KV cache |
 | `PP-DocLayoutV3_fp32_batchable_fp16.onnx`, `PP-OCRv6_{tiny,small,medium}_rec_ctc_fp16.onnx` | FP16 graphs for ONNX Runtime's TensorRT provider | derived from the FP32 file of this repository by `convert_fp16.py` (onnx 1.21.0, onnxconverter-common 1.16.0; I/O stay FP32) |
 
 Pre- and post-processing follow each source model's `inference.yml`.