This is an optimized version based on the original DeepSeek-OCR model, specifically an OCR model that supports inference on Apple Metal Performance Shaders (MPS) and CPU. It can extract text from images and convert it into a structured format, supporting multilingual document recognition.

MultimodalTransformersOther

This is a multi-task fine-tuned model based on DeepSeek-R1-Distill-Qwen-1.5B. It is trained on multiple datasets through the LoRA adapter and has powerful multi-task generalization and reasoning capabilities. It can handle a wide range of natural language and reasoning-based tasks.

Natural Language ProcessingPeftEnglish

This is an image text recognition model based on DeepSeek-OCR, which specifically solves the compatibility issues in the latest version of the transformers library, enabling the model to run smoothly in the latest versions such as transformers v4.57.1.

MultimodalTransformersEnglish

This is a Modal Balance Quantization (MBQ) variant of DeepSeek-OCR. It reduces memory usage by applying 4-bit NF4 quantization to the visual encoder while maintaining the BF16 precision of the projector and language/decoder. It can be locally deployed on ordinary GPUs and used for rapid CPU experiments.

MultimodalTransformers

DeepSeek-OCR-8bit is an MLX format version converted from the DeepSeek-OCR model. It is a visual language model specifically optimized for Apple chips, supporting multilingual OCR recognition and image text understanding tasks.

MultimodalMlxOther

DeepSeek-OCR-6bit is an MLX format version converted from the DeepSeek-OCR model, specifically optimized for Apple chips. This is a vision-language model with powerful optical character recognition capabilities, capable of extracting and recognizing text information from images.

MultimodalMlxOther

This model is a vision-language model in MLX format converted from DeepSeek-OCR, specifically designed for optical character recognition (OCR) tasks. It supports multilingual text recognition and image understanding.

MultimodalMlxOther

This is an MLX format model converted from the DeepSeek-OCR model, supporting multilingual image text recognition and text generation functions, specifically optimized for OCR tasks.

MultimodalMlxOther

DeepSeek-OCR-bnb-4bit-NF4 is a 4-bit NF4 quantized version of deepseek-ai/DeepSeek-OCR, created using bitsandbytes. While maintaining high accuracy, it significantly reduces video memory usage (up to 8GB), making it very suitable for consumer-grade GPUs.

MultimodalSafetensorsOther

This is the MXFP4_MOE quantized version of the DeepSeek-MoE-16B-Chat model, specifically optimized for text generation tasks, providing a more efficient inference solution.

Natural Language ProcessingGguf

DeepSeek-OCR is an advanced optical character recognition model developed by DeepSeek AI, focusing on exploring the boundaries of visual text compression and capable of efficiently extracting and converting text content from images.

MultimodalSafetensorsOther

Qwen-1.5B_THIP is a mathematical reasoning model fine-tuned using the GRPO method on the DeepMath-103k mathematical dataset based on DeepSeek-R1-Distill-Qwen-1.5B with the TRL framework. This model is specifically optimized for solving mathematical problems and has strong mathematical reasoning abilities.

Natural Language ProcessingTransformers

This project performs llama.cpp mixed-precision quantization on DeepSeek-V3-0324, solving the problem of model deployment on edge devices and improving the model's operation efficiency and resource utilization.

Natural Language ProcessingTransformers

This is the Q4_K_M quantized version of the DeepSeek-MoE-16B model from DeepSeek Company. It enables the MoE (Mixture of Experts) CPU offload function through the Rust binding of llama.cpp, significantly reducing video memory usage.

Natural Language ProcessingGgufMultiple Languages

This is a random weight generation model based on the DeepSeek-V3 architecture, specifically designed for testing and development purposes. The model uses randomly initialized weights, and the parameter scale is streamlined, making it suitable for rapid prototype development and functional verification.

Natural Language ProcessingTransformers

DeepSeek-V3.2-Exp-AWQ is a quantized version based on the DeepSeek-V3.2-Exp model, which achieves efficient text generation through the vLLM framework. This model introduces the DeepSeek sparse attention mechanism, significantly improving the efficiency of long context training and inference while maintaining the quality of the model output.

Natural Language ProcessingTransformers

The Transformer Library is a powerful natural language processing tool library that provides a rich set of pre-trained models and convenient usage interfaces. It is widely used in NLP tasks such as text generation, classification, and translation, greatly improving development efficiency.

Natural Language ProcessingTransformers

Mintaka-Qwen3-1.6B-V3.1 is an efficient model focused on scientific reasoning. It is built on Qwen-1.6B and trained on the DeepSeek v3.1 synthetic trajectory (10,000 records). It is optimized for random event simulation, logical problem analysis, and structured scientific reasoning, striking a balance between symbolic precision and lightweight deployment.

Natural Language ProcessingTransformersEnglish

DeepSeek-GRM-16B is a generative reward model based on Self-Principled Critique Tuning (SPCT), which can generate a transparent 'Principle → Critique → Score' evaluation process for query-responses and can be used in tasks such as reinforcement learning, evaluation, and data collection of large language models.

Natural Language ProcessingTransformersMultiple Languages

DeepSeek-V3.1 is a large language model that is post-trained based on DeepSeek-V3.1-Base. It has a 128K context length and supports mixed thinking modes, intelligent tool invocation, and code proxy functions. This model performs excellently in multiple benchmark tests, especially showing significant improvements in mathematical reasoning, code generation, and search agent tasks.

Natural Language ProcessingTransformersMultiple Languages