Brief-details: A pruned 42B parameter variant of Meta's Llama 3 70B, fine-tuned on minipile dataset. Features strong performance on MMLU benchmarks with base model capabilities.
Brief-details: OpenELM-3B is a 3.04B parameter efficient language model from Apple, trained on 1.8T tokens with layer-wise scaling for enhanced accuracy
Brief-details: CodeLlama-13B-Instruct is a powerful code-focused LLM optimized for instruction-following, offering multiple GGUF quantizations for efficient deployment
BRIEF DETAILS: T5-Large model fine-tuned for medical text summarization with 60.5M parameters. Specialized for healthcare documents with Apache 2.0 license.
Brief Details: SmolVLM-Instruct is a 2.25B parameter multimodal model optimized for image-text tasks. Features efficient compression and handles multiple images with impressive performance metrics.
Brief-details: MiniCPM-2B-sft-bf16 is a compact 2.4B parameter LLM that rivals Mistral-7B in performance, supports both English and Chinese, and enables mobile deployment with Int4 quantization.
BRIEF DETAILS: A 7B parameter Mistral-based model enhanced with LASER SVD decomposition, achieving improved reasoning capabilities and higher benchmark scores compared to previous versions.
Brief Details: A conversational AI agent built on DialoGPT-medium that generates personalized responses based on personality profiles and turn-level goals.
Brief Details: NuExtract-large: A 7.39B parameter fine-tuned version of phi-3-small specialized in information extraction, offering precise template-based data extraction capabilities.
Brief Details: A 1B parameter LLM using 1-bit quantization trained on Dolma dataset (60B tokens), demonstrating efficient model compression techniques
Brief-details: A 30B parameter uncensored language model quantized to 4-bit precision, offering multiple GPTQ configurations for different VRAM requirements and performance tradeoffs.
Brief Details: A 70B parameter LLaMA-3 based conversational AI model optimized for creative text generation and roleplay, featuring enhanced prompt adherence and contextual awareness.
Brief-details: DeBERTa-v3-base model fine-tuned on 600+ NLP tasks, optimized for zero-shot classification and NLI with 184M parameters. Apache 2.0 licensed.
Brief-details: A 3B parameter instruction-tuned language model combining Flan and Alpaca training data, offering strong performance for text generation and instruction following.
BRIEF DETAILS: A 7B parameter Mistral-based GGUF model optimized for chat, featuring uncensored responses, multi-turn conversations, and enhanced empathy capabilities. Based on Microsoft's Orca approach.
Brief-details: A specialized LoRA for SDXL 1.0 Base that enhances image quality through negative prompting, featuring improved textures, colors, and anatomical accuracy
Brief Details: WizardMath-70B-V1.0 is a specialized mathematical reasoning LLM achieving 81.6% on GSM8k, surpassing ChatGPT. Built on Llama2, optimized for math.
Brief-details: 12B parameter Mistral-based model optimized for storytelling and roleplay, featuring improved NSFW capabilities and ChatML integration. Trained on diverse datasets with 8K context.
BRIEF-DETAILS: A sophisticated 30B parameter LLM combining multiple LoRAs and model merges, optimized for storytelling and instruction-following with uncensored capabilities
Brief-details: A versatile 8B parameter LLaMA3-based model optimized for roleplay and conversational tasks, featuring enhanced instruction following and balanced SFW/NSFW capabilities.
Brief-details: Pangea-7B is a powerful multilingual, multimodal LLM supporting 39 languages with 7.94B parameters, built on Qwen2-7B-Instruct architecture.
