Brief-details: BART-based conversation summarization model fine-tuned on SAMSum dataset, achieving 54.87 ROUGE-1 score. Popular for dialogue summarization tasks.
BRIEF DETAILS: A specialized anime-style text-to-image diffusion model focused on producing high-quality pastel artwork, featuring unique stylization and detailed character generation capabilities.
Brief Details: FLAN-T5 XXL sharded FP16 model - A powerful text-to-text transformer supporting 50+ languages, optimized for NVIDIA A10G deployment with quantization for efficient inference.
Brief-details: RoBERTa-based ChatGPT detection model trained on HC3 dataset. Achieves text classification for identifying AI-generated content. 53 likes, 4.7K+ downloads.
Brief Details: ImageReward - First general-purpose text-to-image human preference reward model trained on 137k expert comparisons. Outperforms CLIP, Aesthetic, and BLIP.
BRIEF DETAILS: A 2.7B parameter dialogue model fine-tuned from GPT-Neo, designed for conversational AI with advanced text generation capabilities. Features customizable character personas and dialogue formatting.
Brief-details: A specialized text-to-image diffusion model trained on colorized historical photos (1880s-1980s), creating vintage-style images with rich tones using "timeless style" token
Brief-details: FFXIV-Style is a Stable Diffusion model trained on Final Fantasy XIV trailer imagery, specializing in generating game-style character portraits, landscapes, and ornate armor designs.
Brief-details: A Stable Diffusion model fine-tuned on Naruto anime style, enabling text-to-image generation in Naruto aesthetic. Created by Lambda Labs, trained on BLIP-captioned Naruto images.
Brief-details: A Chinese language model with 0.2B parameters, designed for dialogue tasks with comprehensive training pipeline and optimization features.
Brief Details: Chinese text summarization model (523M params) based on PEGASUS, fine-tuned on 7 Chinese datasets. Achieves 48.00 ROUGE-1 score on LCSTS benchmark.
Brief-details: CodeGen-350M-multi is a 350M parameter model for program synthesis, pre-trained on multiple programming languages including Python, Java, and JavaScript. Developed by Salesforce.
Brief-details: Text-to-image diffusion model trained on 8.4M+ images, featuring improved stability, character fidelity, and dual tag/natural language caption support with CCIP metrics surpassing similar models.
Brief-details: A 1.7B parameter multilingual LLM supporting 35 languages, instruction-tuned for European languages with strong MT capabilities & competitive performance.
Brief-details: Llama-3-Chinese-8B-Instruct-v3 is an 8B parameter bilingual (Chinese-English) instruction-tuned LLM, built on Meta's Llama-3 architecture with enhanced conversational abilities.
Brief-details: Multi-lingual translation model (Chinese/Japanese/English) based on MT5, with trimmed vocabulary for efficiency. Supports bidirectional translation between three languages.
Brief-details: A specialized HED ControlNet model designed for FLUX.1-dev, optimized for 1024x1024 resolution image generation with edge detection capabilities.
Brief Details: A bilingual Turkish-English LLM based on LLaMA2 7B, fine-tuned on 10B tokens using LoRA. Optimized for text generation with 6.84B parameters.
Brief Details: A 5.6B parameter text-to-image model developed by a single grad student, featuring T5 integration, MMDiT architecture, and 768x768 resolution capability.
Brief-details: OrpoLlama-3-8B is an 8B parameter fine-tuned LLM based on Meta's Llama-3, optimized with ORPO training and ChatML template, featuring 8k context window.
Brief-details: InternLM2-20B is a powerful 20B parameter language model with 200K token context length, excelling in reasoning, math, and coding tasks, featuring both base and chat-optimized versions.
