This is a model that unexpectedly failed during training due to an excessively small batch size, resulting in a sharp increase in training loss. Interestingly, this model performs excellently in specific tasks such as EWoK, entity tracking, adjective nominalization, COMPS, and AoA, demonstrating the unique phenomenon of 'success in failure'.

Natural Language ProcessingTransformers

A text-to-image model based on Stable Diffusion XL, particularly skilled at generating anime-style illustrations, especially of female characters, with richer skin tone representation and sharper line and shadow effects.

MultimodalDiffusersEnglish

A model for generating stable and sharp normal maps by reducing diffusion variance

Computer VisionDiffusers

Achieves stable and sharp normal effects by reducing diffusion variance

Computer VisionDiffusers

StableNormal is a model that achieves stable and sharp normal map generation by reducing diffusion variance

Computer VisionDiffusers

A diffusion model for generating stable and sharp normal maps by reducing diffusion variance

Computer VisionDiffusers

A zero-shot monocular ranging depth estimation foundation model capable of synthesizing high-resolution depth maps with unparalleled sharpness and high-frequency details

Computer VisionSafetensors

A 90s anime style LoRA specifically designed for Flux models, delivering clean and sharp visual effects

Computer VisionDiffusers

A LoRA model fine-tuned based on FLUX.1-dev, specifically designed to enhance imaging effects, including color consistency, noise reduction, sharpness enhancement, and color vibrancy.

Computer VisionDiffusers

An anime-style model fine-tuned based on Kivotos XL 2.0, specializing in generating Pixiv-style sharp-line anime illustrations

Computer VisionDiffusersEnglish

A model for generating stable and sharp normal maps by reducing diffusion variance

Computer VisionDiffusers

Llama 2 is a pre-trained generative text model with 7 billion parameters released by Meta. It uses an optimized Transformer architecture and is trained on 2 trillion token data. This version is a 7B pre-trained model in Hugging Face format, supporting English text generation tasks.

Natural Language ProcessingTransformersEnglish

LLaMA-13B is a foundational language model developed by Meta AI, featuring 13 billion parameters and based on the Transformer architecture, suitable for large language model research.

Natural Language ProcessingTransformers

wav2vec 2.0 is a self-supervised learning model for speech recognition, pre-trained on a large amount of unlabeled speech data, capable of efficiently performing speech-to-text tasks.

Audio ProcessingTransformers