BRIEF-DETAILS: Adetailer by Kentus is a specialized AI model hosted on HuggingFace, focusing on detail enhancement and refinement in AI-generated content.
Brief Details: InsightFace model releases by hanamizuki-ai - A comprehensive facial recognition and analysis framework deployed on HuggingFace.
BRIEF-DETAILS: Web content categorization model based on gte-base-en-v1.5 (140M params), classifies text into 24 topics without URL dependency. Fine-tuned on Llama-annotated data.
Brief-details: PaliGemma 3B is Google's 3-billion parameter transformer model requiring Hugging Face login for access, focused on advanced language processing tasks.
BRIEF DETAILS: A multilingual sentence embedding model supporting 10 languages, using CoSENT architecture to map sentences to 384-dimensional vectors. Ideal for semantic search and text matching.
Brief-details: MLX-optimized version of DeepSeek-V2.5 language model, converted using mlx-lm v0.18.2. Supports efficient inference on Apple Silicon.
BRIEF DETAILS: Multilingual grapheme-to-phoneme (G2P) conversion model based on ByT5-small architecture, supporting multiple languages for phonetic transcription.
Brief-details: DeepSeek-V3-GGUF is a powerful 671B parameter MoE model with 37B activated parameters, optimized for efficient inference with various quantization options and 128K context length.
Brief Details: Photon_v1 is an AI image generation model by digiplay, available on Hugging Face and Civitai, focused on high-quality photorealistic outputs
Brief-details: A compact variant of InternLM, randomly initialized for experimental purposes. Created by katuni4ka, hosted on HuggingFace for research and development applications.
BRIEF-DETAILS: A 78B parameter multimodal LLM combining InternViT-6B vision encoder with Qwen2.5-72B language model, achieving state-of-the-art performance in visual-language tasks.
Brief-details: CodeGemma-7b is Google's 7B parameter code generation model available on HuggingFace, requiring explicit license acceptance for access
Brief-details: Google's 27B parameter Gemma model optimized for RAG (Retrieval-Augmented Generation) tasks with instruction tuning, requiring license acceptance for access
Brief-details: miniG: A 9B parameter multimodal LLM with 1M context window, trained on 120M synthetic entries. Supports text/image input with focus on high-quality inference over benchmark performance.
Brief Details: MagicAnimate is an advanced AI model for animation tasks, available on HuggingFace. Created by zcxu-eric, it focuses on magical animation generation capabilities.
BRIEF DETAILS: 7B parameter multilingual translation model supporting 10 languages. Specializes in translation tasks, post-editing, and NER. Built on TowerBase with enhanced document-level capabilities.
Brief-details: Collection of specialized YOLO models for face, eye, head/hair and breast segmentation. Features multiple variants trained on custom annotated datasets with different resolutions and targets.
Brief Details: Qwen1.5-4B-Chat is a powerful 4B parameter chat model with 32K context length support, featuring improved multilingual capabilities and significant performance enhancements over its predecessor.
BRIEF DETAILS: A merged SDXL model collection specializing in photorealistic outputs, combining NoobAI and Animagine bases. Features optimized settings for portrait generation with extensive fine-tuning on 20,000+ images.
Brief Details: Unity Sentis 2.1's Whisper-Tiny - A compact speech-to-text model optimized for 16kHz WAV audio transcription in Unity gaming environments
BRIEF-DETAILS: Enhanced RVC v2 pre-trained model offering faster training with minimal data (1 min or even 10s) and reduced epoch requirements for voice conversion
