Brief Details: A 9B parameter multilingual LLM supporting 35 languages, specialized in EU languages. Features GQA architecture and instruction-tuning for enhanced performance.
BRIEF-DETAILS: 4x-Ultrasharp is an AI upscaling model designed for high-quality image enhancement, focusing on sharp detail preservation and 4x resolution scaling.
Brief-details: Anime-focused AI model hosted on HuggingFace, created by a1079602570. Specialized for anime-style content generation and processing, with pruned architecture for optimized performance.
Brief Details: BEN2 (Background Erase Network) - Advanced foreground segmentation model using Confidence Guided Matting for precise background removal and hair matting
Brief Details: LayerDiffusion v1 - An innovative diffusion model by LayerDiffusion team, available on HuggingFace, designed for advanced image generation and processing tasks.
BRIEF DETAILS: Uncensored 14B parameter variant of Qwen2.5-Instruct, fine-tuned with unrestricted dataset. Requires specific system prompt for full functionality.
Brief-details: An uncensored variant of Llama-3.2-11B-Vision-Instruct using abliteration technique, focusing on removing refusal behaviors while maintaining vision capabilities
Brief-details: Advanced audio-to-facial animation transformer model that converts audio into 61 blendshape coefficients for real-time character animation in Unreal Engine
Brief-details: Abliterated version of Qwen2.5-32B-Instruct model utilizing refusal direction techniques, optimized in GGUF format for reduced restrictions
BRIEF-DETAILS: Continuously fine-tuned 14B parameter LLM based on Qwen-2.5, showing improved performance through Ties merge method of base and instruct models.
BRIEF DETAILS: An 11B parameter upscaled version of Mistral-7B with enhanced capabilities. Achieves 65.8% average score on key benchmarks. Uses ChatML format.
Brief-details: VideoCLIP-XL is an advanced video-language model focusing on long description understanding, developed by Alibaba PAI in 2024 with enhanced CLIP capabilities.
Brief Details: ComfyUI_LayerStyle - A specialized extension for ComfyUI enabling advanced layer-based style manipulations in image generation workflows.
BRIEF-DETAILS: Continuous-finetuned 72B parameter LLM based on Qwen2.5, showing strong performance in reasoning tasks with 45.39 avg benchmark score
BRIEF-DETAILS: Fast-SDXL is a specialized collection of LoRA models designed to accelerate noise resolution in Stable Diffusion XL, optimizing inference speed.
BRIEF-DETAILS: State-of-the-art Danish speech transcription model achieving 11.8% WER, fine-tuned from Whisper v3 using Coral & Common Voice datasets. 1.54B parameters.
Brief Details: An uncensored variant of Llama 3.2 11B vision model, featuring multi-modal capabilities and modified behavior for unrestricted responses. Uses 4-bit quantization.
Brief Details: NuExtract-1.5-tiny is a fine-tuned 0.5B parameter model based on Qwen2.5, specialized for structured information extraction with multi-language support and long document handling.
BRIEF-DETAILS: Speech emotion recognition model based on Whisper Large V3, achieving 92% accuracy across 7 emotional states. Trained on RAVDESS, SAVEE, TESS, and URDU datasets.
Brief Details: Llamafiles appears to be a repository focused on Llama model file management and distribution by qjamieq on Hugging Face.
Brief-details: Meta's Llama-3.2-90B-Vision-Instruct is an advanced multimodal AI model combining vision and language capabilities, built on the Llama architecture with 90B parameters.
