Brief Details: A 72B parameter instruction-tuned LLM optimized for coding tasks, built on Qwen2.5-72B-Instruct with improved LiveCodeBench performance scores.
Brief Details: DocOwl2 is an 8.56B parameter multimodal LLM specializing in OCR-free document understanding, featuring efficient page encoding and multi-page processing capabilities.
BRIEF-DETAILS: Hebrew-Mistral-7B: A 7.5B parameter bilingual LLM fine-tuned from Mistral-7B, optimized for Hebrew and English text generation with 64K tokens vocabulary.
Brief-details: A fine-tuned version of Mixtral-8x22B optimized for multilingual conversations with 141B parameters, trained on GPT-4 conversations
Brief Details: 34b-beta is a powerful 34.4B parameter language model optimized for text generation with high MT-Bench scores and low contamination rates.
Brief-details: Qwen1.5-72B-Chat-GGUF is a powerful 72.3B parameter chat model with strong multilingual capabilities and 32K context support, optimized for various quantization levels
BRIEF-DETAILS: A powerful quantization script for converting HuggingFace models to GGUF format with imatrix optimization, ideal for 8GB VRAM NVIDIA GPUs
Brief-details: A 4x13B MoE Llama2-based NSFW-oriented language model utilizing Mixtral architecture, designed for adult content generation with CC-BY-NC-4.0 license
BRIEF DETAILS: LLaMA-Pro-8B-Instruct: Enhanced 8.3B parameter model specialized in programming and math, built on LLaMA2 architecture with advanced reasoning capabilities.
BRIEF DETAILS: Orca-2-13B-GGUF is Microsoft's research-focused 13B parameter model optimized for reasoning tasks, available in various GGUF quantizations for efficient deployment.
Brief-details: Llama-2-13B-GGUF is a GGUF-formatted variant of Meta's 13B parameter LLM, offering multiple quantization options for efficient deployment and strong performance.
Brief-details: A powerful 13B parameter LLM built on Llama2, achieving 91.76% win-rate on AlpacaEval and ranking #1 among 13B models. Excels in text generation and reasoning tasks.
Brief-details: Powerful 70B parameter LLM by Upstage, fine-tuned on Orca-style dataset, achieving top scores on multiple benchmarks with extended context handling capabilities.
Brief-details: A Hugging Face-compatible tokenizer for text-embedding-ada-002, adapted from OpenAI's tiktoken, supporting multiple transformer frameworks
Brief-details: Bilingual Chinese-English 7B parameter LLM fine-tuned on ShareGPT data. Optimized for conversation and coding tasks with support for 4-bit quantization.
Brief-details: ControlNet model for creating artistic QR codes with Stable Diffusion 2.1, enabling generation of scannable QR codes integrated into images
Brief-details: Specialized AI model for generating flat-color ligne claire style anime artwork, featuring delicate lineart and monochrome capabilities. Created by OedoSoldier with CreativeML OpenRail-M license.
Brief-details: A fine-tuned 30B parameter LLaMA model using Alpaca-LoRA methodology and GPT-4 generated data, optimized for instruction-following and chat applications.
Brief-details: BLOOM-1b1 is a 1.07B parameter multilingual language model supporting 46 languages, trained for text generation using an optimized Transformer architecture
BRIEF DETAILS: CodeT5-large: 770M parameter encoder-decoder model for code understanding/generation. Trained on 6 programming languages using masked span prediction.
Brief Details: Minitron-8B-Base is an 8B parameter LLM from NVIDIA, derived from Nemotron-4 15B through efficient pruning and distillation, achieving MMLU score of 64.5.
