Brief Details: Medical Named Entity Recognition model based on DeBERTa-v3, fine-tuned to identify 41 medical entities. 184M params, MIT licensed.
Brief-details: A powerful image segmentation model by Facebook using Mask2Former architecture with Swin backbone, optimized for COCO panoptic segmentation with masked-attention transformer approach.
Brief-details: A 7B parameter instruction-tuned language model by TII, built on Falcon-7B. Apache 2.0 licensed, optimized for chat/instruct tasks with FlashAttention architecture.
Brief-details: GTE-base-en-v1.5 is a 137M parameter English text embedding model supporting 8192 token sequences with SOTA performance on MTEB benchmark and long-context retrieval.
Brief Details: TimeSformer video classification model fine-tuned on Kinetics-600, specialized in space-time attention for video understanding with high-resolution input processing.
BRIEF DETAILS: CodeBERT MLM variant trained on programming languages, built on RoBERTa architecture with 200K+ downloads, specializing in code-language tasks and masked token prediction.
Brief-details: NVIDIA's 70B parameter Llama-3.1 variant optimized for helpful responses, achieving top scores on Arena Hard (85.0), AlpacaEval 2 LC (57.6), and MT-Bench (8.98).
Brief-details: Google's T5-v1.1-XL model - Advanced text-to-text transformer with GEGLU activation, trained on C4 corpus, designed for transfer learning tasks
Brief-details: Advanced text-to-image diffusion model by Playground AI, offering high aesthetic quality at 1024px resolution. Outperforms SDXL and DALL-E 3 in user studies.
Brief Details: ToolACE-8B: State-of-the-art 8B parameter LLM specialized in function calling, based on LLaMA-3.1, achieving GPT-4 level performance on BFCL.
Brief Details: An English-to-German translation model by Helsinki-NLP, achieving BLEU scores up to 45.2 on news test sets. Built on OPUS-MT framework with Marian architecture.
BRIEF DETAILS: A 4-bit quantized version of Codestral-22B using AWQ technology, optimized for efficient text generation with 3.33B parameters. High download count (204K+) suggests strong community adoption.
Brief-details: A specialized CodeBERT model fine-tuned on Java code for 1M steps, designed for code evaluation and masked language modeling tasks.
Brief Details: DialoGPT-medium is Microsoft's state-of-the-art dialogue model trained on 147M Reddit conversations, offering human-like response generation capabilities.
Brief Details: A sentence embedding model that maps text to 768-dimensional vectors, based on DistilRoBERTa with 82.1M parameters. Optimized for semantic similarity tasks.
Brief Details: Vision Transformer base model with 86.9M params, pre-trained on ImageNet-21k and fine-tuned for 384x384 image classification tasks.
Brief Details: A lightweight sentence embedding model (67M params) that maps text to 768-dim vectors, optimized for semantic similarity tasks using TinyBERT architecture.
Brief-details: A 2.61B parameter Gemma-based model fine-tuned for Chinese language tasks, featuring knowledge retrieval capabilities and DPO training, achieving strong performance on various benchmarks.
Brief Details: A powerful 405B parameter LLM optimized for instruction-following, supporting 8 languages and available in GGUF format for efficient local deployment
Brief-details: CrystalClearXL is a popular text-to-image diffusion model with over 214K downloads, built on the StableDiffusionXL pipeline framework offering high-quality image generation capabilities.
Brief-details: A 1.86B parameter vision-language model combining SigLIP and Phi-1.5 architectures, optimized for visual question-answering tasks with competitive performance despite smaller size.
