LFM2-VL-3B is a multimodal vision-language model developed by Liquid AI, built on the LFM2 backbone architecture. It has powerful visual understanding and reasoning capabilities, especially excelling in fine-grained perception tasks. This model can efficiently process text and image inputs and supports native image processing with a resolution of up to 512×512.

MultimodalTransformersMultiple Languages

This project is the result of MXFP4_MOE quantization of the Qwen3-VL-30B-A3B-Thinking model. This model can achieve the conversion from image-text to text, providing a new solution for applications in related fields.

MultimodalGguf

This is a semantic reordering model based on Cross Encoder, specifically fine-tuned for the Redis LangCache semantic caching system. This model can effectively calculate the similarity score of text pairs and is suitable for sentence pair classification and semantic similarity calculation tasks.

Natural Language ProcessingSentence-transformersEnglish

This model is a mixture of experts model after 25% uniform pruning of GLM-4.5-Air using the REAP method. It has been converted to a 4-bit quantized version in the MLX format, suitable for efficient inference on Apple chip devices.

Natural Language ProcessingTransformersEnglish

Zirel-3 is a specific fine-tuned model based on GLM-4.5-Air-REAP-82B-A12B. It is an 82-billion-parameter mixture-of-experts model compressed using the REAP (Routing Weighted Expert Activation Pruning) technology, significantly reducing the model size while maintaining high performance.

Natural Language ProcessingTransformersEnglish

Chandra is an advanced OCR model that can extract text from images and PDFs with high precision and preserve layout information. It supports output in Markdown, HTML, and JSON formats and performs excellently in handwriting recognition, form reconstruction, table processing, etc. It supports more than 40 languages.

MultimodalTransformers

This model is a GGUF format conversion version of cerebras/Qwen3-Coder-REAP-25B-A3B, which uses the llama.cpp tool for format conversion and quantization processing. The original model is a large language model with 25B parameters, specifically optimized for code generation tasks.

Natural Language ProcessingTransformersEnglish

This is a quantized version of the Qwen3-Coder-REAP-25B-A3B model from Cerebras. It uses the imatrix technology of llama.cpp for quantization, which can reduce the model storage space and computational resource requirements while maintaining the model performance.

Natural Language ProcessingGguf

This is the Llamacpp imatrix quantized version of the Ling-flash-2.0 model by inclusionAI. Through advanced quantization technology, it significantly reduces memory usage and computational requirements while maintaining model performance, thereby improving operational efficiency. It supports multiple quantization levels and is suitable for different hardware configurations.

Natural Language ProcessingGguf

This is a quantized version of the Ling-mini-2.0 model from inclusionAI, quantized using the imatrix technology of llama.cpp. It offers various quantization options from 32.54GB BF16 to 4.66GB IQ2_XS and can run efficiently in multiple hardware environments.

Natural Language ProcessingGguf

The INL architecture is a production-level neural architecture based on integral neuron dynamics, replacing traditional feed-forward network layers with iterative dynamics. It is suitable for various types of models such as large language models, vision transformers, and multi-modal models.

Natural Language ProcessingSafetensorsEnglish

GLM-4.5-Air-REAP-82B-A12B is an efficiently compressed version of GLM-4.5-Air. Through the REAP pruning technology, the parameter scale is compressed from 106B to 82B, reducing the memory requirement by 25% while maintaining nearly lossless performance.

Natural Language ProcessingTransformersEnglish

This is a Q4_0 quantized version based on the Qwen3-Next-80B-A3B-Instruct model. Through specific quantization techniques, the efficiency of the model in storage and computation has been significantly improved while maintaining good performance.

Natural Language ProcessingGguf

nanochat-d32 is a lightweight chat model specifically optimized for the ONNX runtime, providing efficient text generation capabilities. This model is converted from karpathy's nanochat-d32 model and supports running in a JavaScript environment.

Natural Language ProcessingTransformers

PixelWave is a text-to-image model developed based on Flux.1. It has been optimized through Nunchaku quantization (SVDQ), providing an efficient image generation solution for users with different GPU configurations.

Computer VisionDiffusersEnglish

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

Qwen3-VL is the most powerful vision-language model in the Tongyi series. It has excellent text understanding and generation capabilities, in-depth visual perception and reasoning capabilities, long context support, enhanced spatial and video understanding capabilities, and strong agent interaction capabilities. This model is the thinking version with 2B parameters, specifically enhanced for reasoning capabilities.

MultimodalTransformers

Qwen3-VL-2B-Instruct-FP8 is the FP8 quantized version of the most powerful vision-language model in the Qwen series. It uses fine-grained fp8 quantization with a block size of 128, and its performance is almost the same as that of the original BF16 model. This model has excellent text understanding and generation capabilities, in-depth visual perception and reasoning abilities, long context support, and enhanced spatial and video dynamic understanding abilities.

MultimodalTransformers

Qwen3-VL is the most powerful vision-language model in the Tongyi series, with comprehensive upgrades in text understanding and generation, visual perception and reasoning, context length, spatial and video understanding capabilities, etc., and it has excellent multimodal interaction capabilities.

MultimodalTransformers

Qwen3-VL-32B-Instruct-FP8 is the FP8 quantized version of the most powerful vision-language model in the Qwen series. It uses fine-grained fp8 quantization with a block size of 128, and its performance metrics are almost the same as the original BF16 model. This model has excellent text understanding and generation capabilities, in-depth visual perception and reasoning abilities, long context support, and strong spatial and video dynamic understanding capabilities.

MultimodalTransformers