MiniCPM-OCR Jev

Use it to screen OCR results for human review or choose the highest-scoring candidate from multiple OCR systems.

Availability

BF16 weights available on Hugging Face and ModelScope

Pricing model

Not established

Deployment

Docker/local SDK, vLLM, SGLang, llama.cpp, Ollama, and FlagOS guides

Privacy

Not established

Licence

Apache License 2.0 for model weights, code, and documentation

• The model returns a score from 0 to 1 for a candidate OCR result, with higher scores indicating greater likelihood of correctness. ^{[1] [2]}
• It assesses text, formula, and table regions. ^[1]

  **MiniCPM-OCR Jev** is a **2B OCR correctness decision model** for text, formula, and table

• BF16 model weights are available from Hugging Face and ModelScope. ^{[1] [2]}

  Model weights, code, and documentation are released under the

• The model has deployment guides for Docker/local SDK, vLLM, SGLang, llama.cpp, Ollama, and FlagOS. ^[1]
• The model can score candidates from different OCR parsers and select the highest-scoring candidate for a region. ^[1]

  scores the candidates for the same region, selects the highest-scoring candidate, and combines

• The model weights, code, and documentation are released under Apache License 2.0. ^[1]

  Model weights, code, and documentation are released under the