BRIEF-DETAILS: Open-source 7B parameter chatbot fine-tuned on ShareGPT conversations, based on LLaMA architecture. Built for research with Apache 2.0 license.
Brief Details: A Chinese T5-based dialogue model trained on billions of tokens for text generation, QA, and creative writing tasks, with enhanced functionality for medical and legal domains.
Brief-details: Flash Diffusion LoRA model that accelerates SD3 to generate 1024x1024 images in just 4 steps, using 90.4M parameters with impressive quality
Brief-details: A Stable Diffusion 1.5-based model trained on 1.2M Instagram images, optimized for Japanese idol/fashion photos with BLIP captioning and booru tags.
Brief Details: Spanish language TTS model based on F5-TTS, trained on 218+ hours of diverse Spanish dialects. Supports multiple regional accents and offers high-quality speech synthesis.
Brief Details: A 6.84B parameter LLaMA2-based chat model optimized for Turkish/English, featuring LoRA fine-tuning on 180K instruction sets.
Brief-details: SDXS-512-0.9 is a real-time one-step latent diffusion model for high-resolution image generation, based on score distillation and feature matching techniques.
Brief Details: TeleChat-7B is a Chinese-English LLM trained on 1.5T tokens with strong performance in knowledge, code, and math tasks. Supports 8K context with 96K extrapolation.
Brief-details: A 13B parameter GPTQ-quantized LLaMA model fine-tuned on ShareGPT, WizardLM, and Wizard-Vicuna datasets, optimized for helpful AI assistance.
Brief Details: 13B parameter LLaMA-based model fine-tuned on ShareGPT and WizardLM datasets, optimized for instruction-following and chat. Now succeeded by Manticore-13B.
Brief Details: 8B parameter code-focused LLM from IBM Research, fine-tuned on permissive instruction data with strong multi-language coding capabilities (57.9% Python pass@1)
Brief Details: AuraFlow-v0.3 is a state-of-the-art flow-based text-to-image model supporting high-resolution outputs up to 1536px, with improved aesthetics and aspect ratio flexibility.
Brief-details: Specialized SD V2-768 embedding model for anthropomorphizing animals with human clothing and abilities. Works best with mammals and birds.
Brief-details: Meta's Llama 3 8B GGUF model - optimized transformer architecture with 8k context window, designed for text generation and dialogue tasks. Latest release: April 2024.
Brief-details: UDOP-large: A 742M parameter universal document processing model for tasks like classification, parsing and visual QA, based on T5 architecture.
Brief-details: Qwen-1.8B-Chat is a 1.8B parameter bilingual chatbot optimized for Chinese/English, featuring 8K context, efficient 2GB inference, and strong performance in reasoning and coding.
Brief Details: Kandinsky-3 is an advanced text-to-image diffusion model featuring an 8.6B text encoder, 3B U-Net, and specialized Russian cultural content generation capabilities.
BRIEF-DETAILS: CodeLlama-7B-GGUF is a 7B parameter code generation model optimized for GGUF format, offering multiple quantization options for efficient deployment and code completion tasks.
Brief Details: OLMoE-1B-7B: Open-source Mixture-of-Experts LLM with 1B active/7B total parameters. State-of-the-art for 1B models, matches Llama2-13B performance.
Brief-details: OpenFlamingo-9B is a deprecated visual-language model based on CLIP and LLaMA-7B, trained on LAION-2B for academic research with promising few-shot capabilities
BRIEF DETAILS: A specialized ControlNet model for SDXL that transforms QR codes into creative, scannable artwork with adjustable parameters for readability and aesthetics.
