CodeV is a 7-billion-parameter vision-language model fine-tuned based on Qwen2.5-VL-7B-Instruct. Through two-stage training of supervised fine-tuning (SFT) and reinforcement learning (RL) based on tool-aware policy optimization (TAPO), it aims to achieve reliable and interpretable visual reasoning. It represents visual tools as executable Python code and ensures the consistency between tool usage and question evidence through a reward mechanism, solving the problem of irrelevant tool invocation under high accuracy.

MultimodalTransformersMultiple Languages

Olmo-3-7B-Instruct-AIO-GGUF is a GGUF quantized version based on the Olmo-3-7B-Instruct model developed by the Allen Institute for AI. This is an autoregressive language model with 7 billion parameters, trained on datasets such as Tulu 2 and UltraFeedback through supervised fine-tuning and direct preference optimization, and performs excellently in question-answering and instruction following.

Natural Language ProcessingTransformersEnglish

Nanbeige 4-3B-Thinking-2511 is the latest enhanced version of the Nanbeige series. Through advanced distillation technology and reinforcement learning optimization, it achieves powerful reasoning ability within a compact 3B parameter scale. This model has achieved state-of-the-art (SOTA) results among models with less than 32B parameters in benchmark tests such as Arena-Hard-V2 and BFCL-V4.

Natural Language ProcessingTransformersMultiple Languages

Olmo 3 is a series of language models developed by the Allen Institute for AI, including two scales of 7B and 32B, with two variants: instructional and reflective. This model performs excellently in long-chain thinking and can effectively improve the performance of reasoning tasks such as mathematics and coding. It adopts a multi-stage training method, including supervised fine-tuning, direct preference optimization, and reinforcement learning with verifiable rewards.

Natural Language ProcessingTransformersEnglish

Olmo-3-7B-Think-DPO is a 7B parameter language model developed by the Allen Institute for AI. It has the ability of long-chain thinking and performs excellently in reasoning tasks such as mathematics and coding. This model has undergone multi-stage training including supervised fine-tuning, direct preference optimization, and reinforcement learning based on verifiable rewards, and is designed specifically for research and educational purposes.

Natural Language ProcessingTransformersEnglish

This is an experimental mixed quantization model that uses the MXFP4_MOE mixed weight technology. While maintaining accuracy close to Q8, it achieves a smaller file size and higher inference speed. The model explores the combination of MXFP4 with high-precision embedding/output weights and achieves near-lossless optimization of accuracy on dense models.

Natural Language ProcessingGgufOther

This model is an MLX format model converted from summykai/gemma3-27b-abliterated-dpo using version 0.28.3 of mlx-lm. It is a 27B parameter Gemma 3 large language model fine-tuned with DPO (Direct Preference Optimization), optimized for efficient operation on Apple Silicon (MLX framework).

Natural Language ProcessingTransformersEnglish

Olmo-3-32B-Think-DPO is a 32B parameter language model developed by Allen AI. It is trained using Direct Preference Optimization (DPO) and has the ability of long-chain thinking reasoning. It performs excellently in complex reasoning tasks such as mathematics and coding.

Natural Language ProcessingTransformersEnglish

This is the llama.cpp quantized version of the Precog-123B-v1 model, which provides a variety of quantization types to meet the needs of different hardware configurations and usage scenarios. This model is a large language model with 123 billion parameters and can run efficiently on various hardware after optimization.

Natural Language ProcessingGguf

Qwen2.5 is a new-generation large language model launched by Alibaba. It has made important optimizations in the chat template, permanently disabled the thinking state, solved problems related to the multi-round chat template, and provided a more convenient experience for experiments and research.

Natural Language ProcessingTransformers

This model is a text generation model fine-tuned based on Meta's Llama-3.1 architecture. It is trained using reinforcement learning with the TRL library and the GRPO (Group Relative Policy Optimization) method, and is specifically optimized for argument generation tasks.

Natural Language ProcessingTransformers

This project is a quantized version based on the IBM Granite-4.0-h-350m model, aiming to make knowledge accessible to everyone. This model is an efficient text generation model. After quantization optimization, it has a smaller model size and faster inference speed.

Natural Language ProcessingGguf

Gaperon-Young-1125-1B is a bilingual (French-English) language model with 1.5 billion parameters, developed by the ALMAnaCH team at the French National Institute for Research in Computer Science and Control (Inria Paris). The model is trained on approximately 3 trillion high-quality tokens, with a particular focus on language quality and general text generation ability rather than benchmark test optimization.

Natural Language ProcessingTransformersMultiple Languages

This is the MXFP4_MOE quantized version of the PromptCoT-2.0-SelfPlay-30B-A3B model, specifically designed for text generation tasks. The model adopts chain of thought prompting and self-play training techniques, and achieves more efficient inference performance through quantization optimization.

Natural Language ProcessingGguf

Qwen3-VL-32B-Instruct is a powerful multimodal vision-language model developed based on the Qwen foundation model. After optimization and quantization, it is specifically optimized for Apple chip devices, providing efficient image-text processing capabilities.

MultimodalTransformers

Based on Stable Diffusion XL, the UNet weights are fine-tuned using the Inversion-DPO method, combining direct preference optimization (DPO) technology and the inversion method to improve the quality and alignment of image generation in the diffusion model.

Computer VisionDiffusers

A-vibe is a Russian-optimized large language model developed by Avito Tech based on Qwen3-8B-Base. Through word segmenter optimization, model slimming, and multi-stage training, it performs excellently in Russian benchmark tests and supports function calls and mathematical problem-solving.

Natural Language ProcessingSafetensorsMultiple Languages

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

Qwen3-14B is a powerful language model that performs excellently in the field of text generation. This model has excellent general capabilities and offers an instruction variant that is friendly to fine-tuning, facilitating the further optimization and application of the model.

Natural Language ProcessingTransformers

Qwen3-4B-SafeRL is a safety-aligned version of the Qwen3-4B model. It enhances the model's robustness against harmful or adversarial prompts through reinforcement learning training. This version uses a hybrid reward function for optimization, balancing three objectives: safety, usefulness, and minimizing unnecessary rejections.

Natural Language ProcessingTransformersEnglish