Brief Details: A lightweight, randomized version of the Marian neural machine translation model, designed for experimental and educational purposes.
Brief Details: Surya Layout is a specialized AI model focused on document layout analysis and understanding, developed by datalab-to for efficient document structure processing.
BRIEF-DETAILS: 24B parameter Mistral model optimized with FP8 quantization, achieving 99.28% accuracy recovery while reducing model size by 50%
Brief-details: A speech diarization model by tezuesh for speaker identification and segmentation in audio conversations, hosted on HuggingFace.
Brief-details: A multimodal vision-language model combining ViT (303M params) with MiniMax-Text-01, featuring dynamic resolution and trained on 512B tokens
Brief-details: OmniAudio-2.6B is a fast, efficient audio-language model combining Gemma-2-2b and Whisper turbo for on-device text/audio processing at 66 tokens/sec.
Brief Details: c4ai-command-r-plus-4bit is a 4-bit quantized language model from CohereForAI, optimized for command-based interactions and efficient deployment with reduced memory footprint.
Brief-details: DeepSeek-V2.5-1210 is an enhanced language model with improved mathematical (82.8% MATH-500) and coding capabilities (34.38% LiveCodebench), featuring BF16 inference support and comprehensive function calling.
BRIEF-DETAILS: ControlNet model specializing in brightness and illumination control for Stable Diffusion, offering precise lighting adjustments with recommended weights of 0.4-0.9
Brief-details: A specialized LoRA model for SD3.5-Turbo focusing on photorealistic image generation, featuring 64 network dimensions and trained on 27 curated images over 13 epochs.
Brief Details: A 12B parameter roleplaying AI model built on Mistral's Nemo base, fine-tuned with hundreds of millions of tokens for creative conversation and character interaction.
BRIEF-DETAILS: Llama 3.2 Vision (11B params) - Advanced multimodal LLM optimized for visual recognition, image reasoning, and captioning tasks
Brief-details: AI Comic Factory is a specialized model by qt8833 focused on comic-style image generation, hosted on HuggingSpace for creative digital artwork and illustration purposes.
Brief-details: Optimized SD3.5 checkpoint with integrated CLIP/text encoders, featuring FP8 precision for efficient deployment in ComfyUI workflows
Brief-details: Abliterated version of Qwen2.5-14B focused on unrestricted responses, featuring 14B parameters and 32K context length with YaRN scaling support
Brief Details: DocLayout-YOLO is a specialized YOLO-based model for document layout analysis, trained on DocStructBench dataset for accurate structure detection.
BRIEF-DETAILS: Modified version of Qwen2.5-32B-Instruct using abliteration technique to reduce safety filters while maintaining core capabilities. 32B parameters.
Brief-details: Fine-tuned 12B parameter model based on Mistral-Nemo-Instruct-2407, designed to replicate Claude 3's prose quality, with 32k context window and GGUF quantization support.
Brief-details: Optimized Whisper large-v3 conversion for CTranslate2, offering faster speech recognition with FP16 precision and seamless integration with faster-whisper framework
Brief-details: Lightweight foundation model for time series classification with 8M parameters. Features easy fine-tuning, dimension reduction adapters, and scikit-learn compatibility.
Brief-details: Moonshine is a lightweight ASR model for real-time speech recognition, featuring tiny (27M) and base (61M) variants optimized for resource-constrained platforms
