BRIEF-DETAILS: 24B parameter Mistral-based model optimized for long-context interactions (up to 24K). Features enhanced vocabulary and detail retention, using Mistral v7 Tekken chat template.
BRIEF-DETAILS: A 24B parameter LLM fine-tuned from Mistral's latest 'Small' model (2501), offering extended context length up to 24K tokens and enhanced detail retention capabilities
Brief-details: Qwen2.5-VL-3B-Instruct-AWQ is a quantized 3B parameter vision-language model optimized for visual understanding, agent tasks, and video analysis with AWQ compression.
Brief Details: OpenThinker-32B is a fine-tuned version of Qwen2.5-32B-Instruct, trained on OpenThoughts-114k dataset with strong performance in mathematical reasoning and problem-solving.
BRIEF-DETAILS: A 7B-parameter multimodal LLM fine-tuned from Qwen2.5-VL, specializing in vision-language understanding and reasoning with 518px image resolution support.
Brief-details: An 8B parameter chat model based on Llama architecture, developed by TAIDE. Features specialized Chinese-English capabilities and community licensing.
BRIEF-DETAILS: 7B parameter LLM fine-tuned on OpenThoughts-114k dataset, achieving strong performance on math and reasoning tasks. Open-source with complete code and data availability.
Brief-details: YuE-s1-7B-anneal-zh-cot is a 7B parameter Chinese music generation model that transforms lyrics into complete songs with vocal and accompaniment tracks under Apache 2.0 license
BRIEF-DETAILS: Pepe LoRA model for image generation with Stable Diffusion, specialized in creating "pepe" style images. Requires "pepe" trigger word in prompts.
Brief-details: A 5B parameter AI safety model by IBM that detects risks in prompts/responses across harm, bias, jailbreaking, and RAG hallucinations with high accuracy
BRIEF DETAILS: A specialized LoRA model for generating images of Korean President Park Jung-hee, designed to work with Stable Diffusion. Triggers with "president_PJH" keyword.
Brief-details: Efficient 500M parameter multimodal model for image+text tasks. Runs on 1.23GB GPU RAM. Features compressed image encoding and optimized architecture.
Brief-details: A specialized LoRA model for generating images of former South Korean President Kim Dae-jung (DJ), designed to work with various Stable Diffusion implementations using the trigger word "president_K_DJ"
BRIEF DETAILS: gte-modernbert-base is a 149M parameter text embedding model with 768-dimensional outputs, supporting 8192 token sequences and achieving 64.38 MTEB score.
Brief Details: A LoRA model for generating images of Korean President Yoon in various scenarios, built for use with Stable Diffusion models and triggered by "president yoon" keyword.
Brief-details: A powerful 70B parameter LLaMA 3.3-based model optimized for storytelling and creative writing, featuring reduced positive bias and enhanced scene descriptions.
Brief-details: AceMath-7B-Instruct is NVIDIA's specialized mathematical reasoning model, fine-tuned from Qwen, achieving 67.2% pass@1 accuracy on math benchmarks.
Brief Details: MiniCPM-o-2_6-gguf is a multimodal AI model optimized for llama.cpp, featuring vision capabilities and GGUF format compatibility with focus on efficient deployment.
Brief Details: A 1B parameter biomedical LLM fine-tuned on 625K examples including 25K chain-of-thought instructions, optimized for healthcare reasoning tasks.
Brief-details: Fine-tuned 8B parameter LLaMa model optimized for tool usage and multi-turn dialogue, achieving SOTA on Berkeley Function-Calling Leaderboard.
Brief-details: A specialized 1.5B parameter translation model built on Qwen2.5, supporting 20+ languages with human-in-the-loop training. Optimized for real-time translation of web content, academic papers, and video subtitles.
