Brief Details: Qwen1.5-110B-Chat is a powerful 111B parameter language model featuring 32K context length, multilingual support, and enhanced chat capabilities without requiring trust_remote_code.
Brief-details: Reader-LM 0.5B is a specialized 494M parameter model for HTML-to-Markdown conversion with 256K context length, built on Qwen architecture and supports multilingual content.
Brief-details: Bark-voice-cloning is a feature extraction model for voice cloning and speech transfer, utilizing HuBERT outputs to generate bark-compatible semantic tokens.
Brief Details: AnimateDiff-A1111 is a comprehensive collection of animation models for Stable Diffusion WebUI, featuring motion modules, LoRAs, and domain adapters in fp16 format.
Brief-details: A powerful 34B parameter bilingual Chinese-English LLM built on Yi-34B, featuring 8K context window and state-of-the-art performance in benchmarks.
Brief Details: DiscoLM-mixtral-8x7b-v2: A 46.7B parameter MoE model based on Mistral AI's Mixtral architecture, fine-tuned on Synthia, MetaMathQA, and Capybara datasets.
Brief-details: An improved anime latent diffusion model featuring enhanced dark rendering, bioluminescence effects, and VAE tuning. Built on SomethingV2 with merged model capabilities.
Brief-details: A 30B parameter uncensored language model optimized for CPU+GPU inference, available in multiple GGML quantization formats with sizes ranging from 13.6GB to 34.56GB
BRIEF-DETAILS: End-to-end OCR model (4.25B params) handling text, math latex & markdown formats. Runs in 4-bit with 2.8GB VRAM. Based on Phi-3.5-vision-instruct.
Brief-details: An 80M parameter BERT variant optimized for long-context retrieval (32k sequence length), featuring Monarch Mixer architecture for efficient processing.
Brief-details: TemporalNet2 is an advanced ControlNet model enhancing temporal consistency in video generation using frame and optical flow guidance
Brief-details: Falcon-180B-Chat-GGUF is a powerful 180B parameter chat model optimized for inference, featuring multi-query attention and supporting multiple languages with GGUF quantization formats.
Brief-details: An advanced 9.2B parameter language model fine-tuned using SimPO (Simple Preference Optimization) on Google's Gemma architecture, optimized for enhanced conversational AI capabilities
BRIEF-DETAILS: A powerful 30B parameter LLaMA model merged with SuperCOT-LoRA, optimized for chain-of-thought reasoning and langchain integration
BRIEF DETAILS: FireFunction-v1: A 46.7B parameter function-calling model offering GPT-4 level quality with 4x faster inference. Optimized for structured data extraction and routing.
Brief-details: MPT-7B-StoryWriter optimized for 4-bit quantization - 1.07B params storytelling model with 65k+ context length using ALiBi attention
Brief-details: WizardLM 7B uncensored GGML - Quantized model variants (2-8 bit) optimized for CPU+GPU inference, based on WizardLM without alignment constraints
BRIEF DETAILS: PEGASUS-based financial news summarizer with 569M params, fine-tuned on Bloomberg articles. Achieves 23.55 ROUGE-1 score for concise financial summaries.
Brief-details: MOSS is a 16B parameter multilingual LLM from Fudan University with plugin support, capable of natural conversations in English and Chinese with extensive safety controls
Brief-details: MidJourney-PaperCut is a specialized text-to-image model trained on v1-5 base for 7000 steps, requiring minimal prompt engineering with the 'mdjrny-pprct' token
Brief-details: A specialized DreamBooth model fine-tuned for generating flat, diffuse textures with minimal lighting/shadows, based on Stable Diffusion 2 and trained on CC0 PolyHaven textures.
