Brief-details: A 4B parameter Qwen2.5 MOE model combining censored and uncensored DeepSeek variants, optimized for reasoning with 128k context.
Brief-details: BioBERT is a biomedical language model pre-trained on large-scale biomedical corpora, based on BERT architecture. Optimized for biomedical text mining tasks and NLP applications in healthcare.
Brief-details: Marigold Normals v0-1: A deprecated fine-tuned diffusion model for monocular surface normals estimation, supporting 768px resolution with DDIM scheduler
Brief Details: Portuguese pun recognition model achieving 69% F1-Score, fine-tuned on Puntuguese dataset. Supports text classification for humor detection.
Brief Details: BM25 ranking model by Qdrant for relevance-based document search, optimized for FastEmbed integration with sparse vector support and IDF modification.
BRIEF-DETAILS: LongT5 with transient-global attention mechanism - specialized transformer for processing long sequences up to 16K tokens, ideal for summarization and QA tasks
BRIEF-DETAILS: Qwen2.5's 32B quantized instruction model offering 131K context, multi-language support, and enhanced capabilities in coding, math, and long-text generation.
Brief Details: Audio Spectrogram Transformer fine-tuned on AudioSet - Converting audio to spectrograms for classification using Vision Transformer architecture
BRIEF-DETAILS: 32B parameter LLaMA-based model with multiple GGUF quantizations (Q2-Q8) for efficient deployment, featuring specialized formats for ARM/AVX systems
Brief-details: EVA Giant vision model with 1B params, trained on 30M images. Features 560x560 input size and achieves 89.79% top-1 accuracy on ImageNet-1k.
Brief-details: NVIDIA's Cosmos-1.0-Diffusion-7B-Text2World is a 7B parameter diffusion model designed for text-to-world generation, leveraging advanced AI capabilities for environmental understanding and generation.
BRIEF-DETAILS: NVIDIA's 7B parameter diffusion model designed for video understanding and world modeling, focusing on video-to-world generation tasks.
BRIEF-DETAILS: A specialized reading order model called surya_order by vikp, designed for determining optimal reading sequences in surya applications.
Brief Details: LLaMA-7B is a powerful 7-billion parameter language model developed by Meta AI, hosted by nyanko7 on HuggingFace, optimized for efficient natural language processing.
Brief-details: MiniCPM-Llama3-V 2.5 is a GGUF-formatted language model optimized for llama.cpp, focusing on efficient deployment and compatibility with Ollama framework.
BRIEF-DETAILS: VoiceCraft is a text-to-speech synthesis model developed by pyp1, focusing on high-quality voice generation and manipulation with neural networks
Brief-details: CodeGemma-7B-IT is Google's 7B parameter code-focused language model requiring Hugging Face authentication and license agreement, specializing in code generation and understanding tasks.
BRIEF-DETAILS: DeepSeek-R1-Distill-Qwen-7B: A 7B parameter distilled model from DeepSeek-R1, optimized for reasoning tasks with strong math and coding capabilities
Brief-details: Specialized 2B parameter model for structured information extraction, supporting multimodal inputs and multilingual capabilities based on InternVL2.5
BRIEF DETAILS: A specialized embedding model for single-cell RNA sequencing data that converts gene expression patterns into high-dimensional vectors for advanced biological analysis.
Brief-details: EliGen is an advanced entity-level controlled image generation model that enables precise control over specific image regions through regional attention mechanisms and supports multiple generation modes including inpainting and style transfer.
