BRIEF DETAILS: Vicuna 13B unfiltered language model trained on V2023.05.02v0 dataset, featuring unrestricted responses and multiple quantization formats for deployment flexibility.
Brief-details: Vistral-7B-Chat is a 7B parameter chat model from Viet-Mistral, focused on safe and ethical AI interactions with specific guidance against harmful human experimentation.
Brief-details: Function-calling enabled Llama 2 7B model optimized for structured API interactions. Supports multiple function calls and argument types with improved v2 syntax.
BRIEF-DETAILS: Alpaca LoRA 30B GGML - Optimized version of Alpaca fine-tuned LLaMA for CPU inference, compatible with Alpaca.cpp and related frameworks.
Brief Details: YaLM-100B is Yandex's 100B parameter GPT-like model trained on 1.7TB of multilingual data, optimized for English and Russian text generation and processing.
Brief-details: Pyannote embedding model for speaker diarization and voice processing tasks. Supports academic research and commercial applications with focus on machine listening.
Brief-details: A GGUF-optimized version of OLMoE-1B-7B-0125-Instruct model by Allen AI, combining mixture-of-experts architecture with instruction tuning capabilities
Brief Details: BERT model fine-tuned for financial transaction categorization across 25 categories, optimized for English language processing and classification tasks.
Brief-details: Qwen 2.5 7B-based model fine-tuned on GammaCorpus v2-100k dataset, featuring 7.61B parameters and trained for 60 epochs on T4 GPU. Optimized for chat.
BRIEF-DETAILS: FLUX.1-Depth-dev-onnx is a depth estimation model in ONNX format from black-forest-labs, designed for non-commercial applications with specialized depth perception capabilities.
Brief-details: Compact 3B-parameter vision-language model optimized for document understanding, featuring strong performance on chart/table analysis and general VQA tasks. Built on Granite LLM.
Brief-details: Quantized version of Selene-1-Mini-Llama (8B params) offering multiple GGUF compression variants, optimized for different size/quality tradeoffs, with Q4_K_M recommended for balanced performance.
Brief-details: DeepSeek-R1 quantized to INT4 with symmetric quantization and GGUF format, optimized using Intel's auto-round algorithm for efficient inference while maintaining performance.
Brief-details: Blossom-V6-14B is an open-source conversational LLM based on Qwen2.5-14B, featuring innovative data synthesis workflow and cross-model evaluation for enhanced performance.
Brief-details: A GGUF-formatted Japanese language model converted from CyberAgent's DeepSeek-R1-Distill-Qwen-14B, optimized for Japanese text generation and processing
Brief Details: 2B parameter multimodal model fine-tuned on 8k curated dataset using GRPO, supporting English/Chinese vision-language tasks with efficient processing capabilities.
Brief-details: A 1B parameter LLaMA model fine-tuned on GSM8K dataset for mathematical reasoning, trained over 132 steps by NickyNicky. Available on HuggingFace.
Brief Details: RouWei-0.7 is a large-scale anime art model fine-tuned from Illustrious, featuring 7M unique images, enhanced prompt following, and superior anatomy rendering.
Brief Details: A 70B parameter LLaMA-based model combining EVA-LLAMA storytelling, EURYALE scene descriptions, and DeepSeek-R1 reasoning, optimized for creative dialogue and detailed narratives.
BRIEF DETAILS: A comprehensive GGUF quantized variant of the sororicide-12B model, offering multiple compression levels from 3.1GB to 10.2GB with imatrix optimizations
BRIEF-DETAILS: DeepSeek-R1-Distill-Qwen-14B quantized model with multiple compression options (Q2-Q8), optimized for efficient deployment and reduced size.
