Qwen3-VL-TimeTravel is a version fine-tuned on the MBZUAI/TimeTravel dataset using the Unsloth library based on the Qwen3-VL-8B-Instruct model. This model is specifically designed to generate descriptions of historical cultural relic images and has professional capabilities in historical and cultural relic analysis.

MultimodalTransformersEnglish

This is a semantic re-ranking model based on the Cross Encoder architecture, specifically fine-tuned for the Redis LangCache semantic caching scenario. The model is trained on the LangCache Sentence Pairs dataset and can effectively calculate the semantic similarity scores of text pairs for sentence pair classification and re-ranking tasks.

Natural Language ProcessingSentence-transformersEnglish

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

LibreFLUX-ControlNet is a text-to-image generation model based on the ControlNet architecture, using LibreFLUX as the base Transformer model. This model is trained on the SA1B dataset and can generate high-quality image content based on text prompts and control images.

Computer VisionDiffusers

BEVANet is a deep learning model designed for real-time semantic segmentation, which performs excellently on datasets such as Cityscapes. It achieves an outstanding performance of 81.0% mIoU and 32.8 FPS on the RTX3090, balancing the requirements of accuracy and speed.

Computer VisionSafetensors

This is a quantized version of the Beck-4B model by gustavecortal. It uses the llama.cpp tool and the imatrix dataset for quantization optimization, providing a variety of quantization types from BF16 to IQ2_M, and can run efficiently in multiple hardware environments.

Natural Language ProcessingGguf

This is the quantized version of the YanoljaNEXT-Rosetta-12B-2510 model from Yanolja. It is quantized using the llama.cpp tool and the imatrix dataset, offering multiple quantization types to adapt to different hardware environments, significantly reducing resource consumption while maintaining model quality.

Natural Language ProcessingGguf

Stable Diffusion v1-5 is a text-to-image generation model based on latent diffusion technology, capable of generating realistic images according to text descriptions. This model is trained on the LAION-2B dataset, supports English text input, and generates images with a resolution of 512x512.

MultimodalDiffusers

Tiny-Purr-350M is a conversational model fine-tuned based on LiquidAI/LFM2-350M. It is trained on the Tiny-Purr-2 dataset and has the ability to provide casual, friendly, and cat-themed conversational responses, supporting both Chinese and English.

Natural Language ProcessingPeftMultiple Languages

This is a quantized version of swiss-ai's Apertus-8B-Instruct-2509 model. The llama.cpp tool and the imatrix dataset are used to generate GGUF files of various quantization types, which are convenient for efficient operation in different hardware environments.

Natural Language ProcessingGguf

Wan2.2 is a significant upgrade version of the basic video model. It introduces the Mixture of Experts (MoE) architecture, incorporates carefully curated aesthetic data, and is trained on larger datasets to enhance the ability to generate complex motions. This model supports generating 5-second videos with 480P and 720P resolutions, and has significant improvements in video generation quality and performance.

MultimodalDiffusers

This is a Vision Transformer model based on RoPE (Rotary Position Encoding) with a parameter scale of 150M and a patch size of 14x14. The model has been pre-trained on a diverse dataset of approximately 21 million images and can be used as a general visual feature extractor.

Computer VisionPytorch

This is the GGUF quantized version of the IBM Granite-4.0-h-micro model. The quantization process is completed using the llama.cpp tool in combination with a specific dataset, providing multiple quantization type files for efficient operation under different hardware conditions.

Natural Language ProcessingGguf

Granite-4.0-H-Small is a long-context instruction model developed by IBM with 32 billion parameters, fine-tuned based on Granite-4.0-H-Small-Base. This model combines open-source instruction datasets and internal synthetic datasets, and uses techniques such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has significantly improved instruction following and tool invocation capabilities, and is particularly suitable for enterprise-level applications.

Natural Language ProcessingTransformers

Granite-4.0-H-Micro is a 3-billion parameter long-context instruction model developed by IBM, fine-tuned from Granite-4.0-H-Micro-Base. This model combines open-source instruction datasets and internal synthetic datasets, and is developed using techniques such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has a structured chat format and performs excellently in instruction following and tool invocation capabilities.

Natural Language ProcessingTransformers

Granite-4.0-H-Tiny is a long-context instruction model developed by IBM with 7 billion parameters, fine-tuned based on Granite-4.0-H-Tiny-Base. This model combines open-source instruction datasets and internal synthetic datasets, and is developed using techniques such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has enhanced instruction-following and tool invocation capabilities, making it particularly suitable for enterprise-level applications.

Natural Language ProcessingTransformers

Granite-4.0-Micro is a long-context instruction model with 3 billion parameters developed by IBM, fine-tuned based on Granite-4.0-Micro-Base. This model combines open-source instruction datasets and internal synthetic datasets, and is developed using technologies such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has enhanced instruction-following and tool invocation capabilities, and is particularly suitable for enterprise-level applications.

Natural Language ProcessingTransformers

Granite-4.0-H-Micro is a 3-billion-parameter long-context instruction model developed by IBM, fine-tuned from Granite-4.0-H-Micro-Base. This model is trained by combining an open-source instruction dataset and an internal synthetic dataset, and it has enhanced instruction-following and tool call capabilities, making it particularly suitable for enterprise-level applications.

Natural Language ProcessingTransformers

A version fine-tuned based on google/medgemma-4b-it, specifically designed for binary breast X-ray image classification on the OMAMA 256×256 dataset. This model uses LoRA (Low-Rank Adaptation) for parameter-efficient adjustment and performs excellently in the breast cancer classification task, with the accuracy increasing from 12% of the base model to 95%.

MultimodalTransformersEnglish

This is a vision transformer model pre-trained on DINOv2, specifically optimized for retinal color fundus photos. The model is trained on a subset of the MEH AlzEye dataset and is provided in a format compatible with Hugging Face Transformers, facilitating retinal image feature extraction and downstream medical imaging tasks.

Computer VisionTransformersOther