Indic OCR MLX Exports
Collection
MLX Exports of Bodahn AI's Indic OCR for different precisions • 3 items • Updated
You need to agree to share your contact information to access this model
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
Please provide your details and agree to the LICENSE [simpler version] to request access.
Log in or Sign Up to review the conditions and access this model content.
IndicOCR MLX BF16 (IndicBlockOCR)
This repository provides the full-precision BF16 (Bfloat16) reference MLX export of IndicBlockOCR (Stage 2 of bodhan-ai/indic-ocr), optimized for Apple Silicon Metal GPUs (M1/M2/M3/M4).
The recognizer architecture is based on Qwen3.5-0.8B with a 151k-token Sarvam-30B vocabulary covering 22 official Indic scripts, English, numbers, and mathematical notation.
Precision Variants
hari31416/indic-ocr-mlx-4bit: 636 MB model weights, 391 ms/crop (24.96x speedup), 4.83% CER.hari31416/indic-ocr-mlx-8bit: 1.0 GB model weights, 486 ms/crop (20.09x speedup), 1.56% CER.hari31416/indic-ocr-mlx-bf16(This repository): 1.6 GB model weights, 683 ms/crop (14.30x speedup), 1.69% CER.
Empirical Parity and Accuracy Benchmarks
Evaluated against the PyTorch reference baseline on 50 real-world Indic document crops from ai4bharat/indicdlp.
Stage 2 Recognition Parity (BF16 vs PyTorch Baseline)
| Metric | PyTorch Baseline (CPU) | IndicOCR MLX BF16 (Metal GPU) | Parity Delta / Speedup |
|---|---|---|---|
| Character Error Rate (CER) | Reference (0.0) | 0.0169 (1.69%) | Reference fidelity |
| Word Error Rate (WER) | Reference (0.0) | 0.0410 (4.10%) | Reference readability |
| Exact Match Rate | 100.0% | 72.0% | See analysis below |
| Crops with CER <= 5% | 100.0% | 94.0% | High fidelity |
| Crops with CER <= 10% | 100.0% | 94.0% | Production-ready |
| Crops with CER <= 20% | 100.0% | 98.0% | Production-ready |
| Latency per Crop | 9,768.58 ms | 683.05 ms | 14.30x Speedup |
| Throughput | 15.3 chars/sec | 218.9 chars/sec | 14.3x Throughput |
| Memory Footprint | ~1.7 GB | ~1.6 GB | Reference size Reduction |
Stage 1 Layout Detection (PP-DocLayoutV3 on PyTorch MPS)
Stage 1 runs with PyTorch on Apple's MPS backend. The recognizer uses MLX, but the layout detector does not: this project has no MLX PP-DocLayoutV3 runtime, and converting the layout tensors alone does not improve inference.
| Metric | Score | Note |
|---|---|---|
| Mean IoU | 0.9983 | Near-perfect bounding box alignment |
| Label Match Rate | 100.0% | 37 document class categories identical |
| Reading Order Match Rate | 100.0% | Pairwise reading order identical |
| Detection Latency | 156.61 ms | 3.85x Speedup vs PyTorch CPU (603 ms) |
Understanding Exact Match vs Character Error Rate
While the Exact Match rate is 72.0%, the mean Character Error Rate is only 1.69% and 94.0% of crops have less than 5% CER. Exact Match is an all-or-nothing metric: on long paragraph crops (e.g. 600+ characters), a single whitespace difference or character substitution drops the entire crop score to 0.
Autoregressive vision-language decoding on Apple Silicon Metal GPU uses different matrix multiplication tiling, reduction order, and floating-point accumulation compared to NVIDIA CUDA kernels, causing tiny logit variations (~1e-4) that can flip argmax selections at near-tie probability boundaries.
Quickstart
1. Installation
pip install -r requirements.txt
2. Interactive Gradio Web Studio
Launch the visual interface with interactive bounding box hover tooltips, reading-ordered Markdown preview, and JSON inspection:
python app.py
Then open http://127.0.0.1:7860 in your browser.
3. End-to-End Python Pipeline
Run the full two-stage document pipeline (Stage 1 Layout Detection + Stage 2 Text Transcription):
from pipeline import IndicOCRPipeline
# Initialize end-to-end pipeline (auto-loads Stage 1 and Stage 2 on Apple Silicon)
ocr = IndicOCRPipeline()
# Process full document page
result = ocr.process("path/to/document.png")
# Reading-ordered Markdown
print(result.markdown)
# Access individual detected blocks
for block in result.blocks:
print(f"#{block.order} [{block.label}] {block.bbox_xyxy}: {block.text}")
4. Single Crop CLI Transcription
Quickly transcribe a specific cropped text line or region:
python transcribe.py --image path/to/crop.png
5. Direct mlx_vlm Usage
Load the recognizer directly using standard MLX VLM:
from mlx_vlm import generate, load
from PIL import Image
model, processor = load("hari31416/indic-ocr-mlx-bf16")
prompt = processor.apply_chat_template(
[
{
"role": "user",
"content": [
{"type": "image"},
{"type": "text", "text": "Transcribe the text in this image."},
],
}
],
add_generation_prompt=True,
tokenize=False,
)
image = Image.open("crop.png").convert("RGB")
output = generate(model, processor, prompt=prompt, image=[image], max_tokens=256)
print(output.text.strip())
Repository Structure
├── README.md # Model card and parity benchmarks
├── app.py # Interactive Gradio web application
├── pipeline.py # Importable end-to-end Python pipeline
├── transcribe.py # Single-crop CLI transcription script
├── config.json # Recognizer model configuration
├── generation_config.json # Generation hyperparameter defaults
├── model.safetensors # BF16 recognizer weights (~1.6 GB)
├── model.safetensors.index.json # Weight index mapping
├── processor_config.json # Vision & image processor configuration
├── tokenizer.json # Sarvam-30B tokenizer
├── tokenizer_config.json
├── chat_template.jinja # Multimodal chat template
├── requirements.txt # Minimal dependencies
├── .gitattributes # Git LFS tracking configuration
└── layout/ # Legacy experimental tensor export, not used at runtime
├── config.json
└── model.safetensors # PP-DocLayoutV3 tensors; requires an MLX model implementation
License and Citations
This model is distributed under the Indic Open Model License v1.0. See the Indic Open Model License and the Plain-Language License Deed for complete terms.
If using this export, please cite the original Bodhan AI project:
@misc{bodhan2026indicocr,
author = {Bodhan AI},
title = {IndicOCR: High-Performance Document OCR for Indic Languages},
year = {2026},
url = {https://huggingface.co/bodhan-ai/indic-ocr}
}
- Downloads last month
- -
Model size
0.9B params
Tensor type
BF16
·
Hardware compatibility
Log In to add your hardware
Quantized
Model tree for hari31416/indic-ocr-mlx-bf16
Base model
bodhan-ai/indic-ocr