Image-to-Text
MLX
Safetensors
mlx-weights
paddlepaddle-ocr
ppocrv5
ppocrv6
ppdoclayoutv3
pp-structure
apple-silicon
Instructions to use plaincompute/ppocr-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use plaincompute/ppocr-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir ppocr-mlx plaincompute/ppocr-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
| { | |
| "model_type": "pp_ocrv6_medium_det", | |
| "mode": "large", | |
| "upsample_mode": "nearest", | |
| "upsample_align_mode": 1, | |
| "backbone_config": { | |
| "model_type": "pp_lcnet_v4", | |
| "stem_channels": [3, 64, 128], | |
| "stem_type": "large", | |
| "out_features": ["stage1", "stage2", "stage3", "stage4"], | |
| "out_indices": [1, 2, 3, 4], | |
| "block_configs": [ | |
| [[3, 128, 128, 1, true], [3, 128, 128, 1, false]], | |
| [ | |
| [3, 128, 256, 2, false], | |
| [3, 256, 256, 1, true], | |
| [3, 256, 256, 1, false] | |
| ], | |
| [ | |
| [3, 256, 512, 2, false], | |
| [3, 512, 512, 1, true], | |
| [3, 512, 512, 1, false], | |
| [3, 512, 512, 1, true], | |
| [3, 512, 512, 1, false] | |
| ], | |
| [ | |
| [3, 512, 896, 2, false], | |
| [3, 896, 896, 1, true], | |
| [3, 896, 896, 1, false] | |
| ] | |
| ] | |
| }, | |
| "use_lab": false, | |
| "use_last_conv": true, | |
| "class_expand": 2048, | |
| "class_num": 1000, | |
| "out_indices": [0, 1, 2, 3], | |
| "neck_out_channels": 256, | |
| "reduce_factor": 2, | |
| "intraclass_block_number": 4, | |
| "intraclass_block_config": { | |
| "reduce_channel": [1, 1, 0], | |
| "return_channel": [1, 1, 0], | |
| "vertical_long_to_small_conv_longratio": [[7, 1], [1, 1], [3, 0]], | |
| "vertical_long_to_small_conv_midratio": [[5, 1], [1, 1], [2, 0]], | |
| "vertical_long_to_small_conv_shortratio": [[3, 1], [1, 1], [1, 0]], | |
| "horizontal_small_to_long_conv_longratio": [[1, 7], [1, 1], [0, 3]], | |
| "horizontal_small_to_long_conv_midratio": [[1, 5], [1, 1], [0, 2]], | |
| "horizontal_small_to_long_conv_shortratio": [[1, 3], [1, 1], [0, 1]], | |
| "symmetric_conv_long_longratio": [[7, 7], [1, 1], [3, 3]], | |
| "symmetric_conv_long_midratio": [[5, 5], [1, 1], [2, 2]], | |
| "symmetric_conv_long_shortratio": [[3, 3], [1, 1], [1, 1]] | |
| }, | |
| "head_in_channels": 1024, | |
| "scale_factor": 2, | |
| "scale_factor_list": [1, 2, 4, 8], | |
| "hidden_act": "relu", | |
| "kernel_list": [3, 2, 2] | |
| } |