BRIEF DETAILS: 7B parameter creative writing model built on LLaMA 3.2, optimized for fiction/storytelling with 128k context. Features enhanced prose generation and uncensored output capabilities.
Brief-details: GLiREL is a zero-shot relation extraction model that can classify relationships between entities in text without prior training on specific relation types.
Brief-details: A specialized LoRA model trained on FLUX.1-dev for generating professional logo designs. Features 64 network dimensions, trained on 14 hi-res images over 10 epochs.
Brief Details: AIMv2-1B: 1.23B parameter vision model by Apple achieving 88.1% ImageNet accuracy. Excels in multimodal tasks and feature extraction with PyTorch/JAX support.
Brief Details: A specialized LoRA model for FLUX.1-dev focused on generating realistic 3D polygon character portraits with high detail and customizable features. 64 network dimensions, AdamW optimizer.
Brief Details: A specialized LoRA fine-tuning of Stable Diffusion 3.5 focused on photorealistic image generation, trained on 40 curated images with optimized parameters for natural results.
Brief Details: A 2.18B parameter vision-language model combining Qwen2.5-1.5B-instruct LLM with SigLIP vision architecture, trained on 40M image-text pairs for multimodal understanding.
Brief Details: Llama-3.2-1B-Instruct is a lightweight ONNX-compatible language model optimized for web deployment via Transformers.js, featuring conversational abilities.
BRIEF DETAILS: Specialized Russian ASR model fine-tuned from Whisper Large V3, achieving 6.39 WER on Common Voice dataset. 1.54B params, optimized for phone calls.
Brief-details: Multilingual cross-encoder model trained on MMARCO dataset, supporting 15 languages for text classification and information retrieval tasks
Brief-details: A specialized 72B parameter language model fine-tuned for roleplay and creative writing, built on Qwen2.5-72B with advanced instruction-following capabilities
Brief Details: A 7B parameter LLM built on Qwen2.5, specializing in visual reasoning and summarization with 32K context window and multi-agent capabilities
Brief Details: A specialized LoRA model for anime-style image generation built on FLUX.1-dev. Features 64 network dimensions, trained on 50 high-res images with "Animeo" trigger word.
Brief Details: Russian language Mistral-based 12B parameter LLM fine-tuned for dialogue and instruction following. Optimized for BF16 precision with strong performance on RuArenaHard benchmarks.
Brief Details: A powerful 13B parameter Japanese-English language model with instruction tuning, optimized for conversational AI and text generation tasks. Excels in both languages with extensive pre-training.
BRIEF DETAILS: Fine-tuned Whisper model optimized for Chinese ASR with 24-64% performance improvement over base model. Supports high-accuracy speech recognition with integrated punctuation.
Brief-details: Open-source medical image embedding model enabling zero-shot classification and embeddings for medical imaging, with simplified implementation and FastAPI integration
Brief Details: A specialized LoRA collection for FLUX.1-dev model, featuring high-quality character generation models like Black Myth: Wukong. Non-commercial license, optimized for text-to-image generation.
Brief Details: LoRA model for FLUX.1-dev that creates early 2000s animation-style images, optimized at 1.2 strength with natural prompting support.
Brief Details: F-Meme is a FLUX-based LoRA model specialized in meme image generation, offering 64 network dimensions and trained on 10 curated images
Brief-details: CogVideoX LoRA model trained on Arcane style, enables text-to-video generation with anime-style aesthetics. Non-commercial research project with character tokens.
