GLM 4.6 is a quantized model optimized for a configuration of 128GB memory + single GPU. It uses the IQ_K quantization method, providing better quality and performance than the mainstream llama.cpp at the same size. This model needs to be used with ik_llama.cpp. On a configuration of 128GB dual-channel DDR5 memory, a single CCD Ryzen 7000 processor + a single 3090 graphics card, the text generation speed can reach approximately 6.8 tokens per second.

Natural Language ProcessingGgufMultiple Languages

This is a Japanese pre - trained language model based on the Mistral 300M architecture. It is trained using the Wikipedia and cc100 datasets, and the byte - fallback technique in the SentencePiece tokenizer is adopted to suppress the generation of unknown words.

Natural Language ProcessingTransformers

Pola Photo (Flux Dev) is an image generation model based on Flux.1-dev, specifically designed to generate realistic images with the style of Polaroid instant photos. This model is trained on the Replicate platform using open-source materials and follows the CC non-commercial attribution license agreement.

Computer VisionDiffusers

AiGirl_II is a text-to-image generation model built on black-forest-labs/FLUX.1-dev, combining LoRA technology and the Diffusers library, specifically designed for generating images in a specific style. This model uses the CC BY-NC-ND 4.0 license and is suitable for non-commercial use.

Computer VisionDiffusers

A free text-to-image generation model built on Stable Diffusion 1.5. Fine-tuned through multiple LoRA modules, the generated images follow the CC0 license and can be used commercially for free without attribution.

Computer VisionDiffusersEnglish

Video classification model fine-tuned on a subset of UCF101 based on the VideoMAE base model

Computer VisionTransformers

OuteTTS-0.3-500M is a multilingual text-to-speech model developed by OuteAI and released under the cc-by-nc-4.0 license.

Audio ProcessingGgufMultiple Languages

A neural vocoder based on BigVGAN, trained with specific mel-spectrogram inputs, suitable for high-quality audio generation tasks

A Japanese/English text-to-video generation model trained on CC-BY and CC-0 licensed images, developed with support from Japan's Ministry of Economy, Trade and Industry GENIAC program

MultimodalDiffusersMultiple Languages

CLIP-style image encoder based on ResNet50 architecture, trained on CC12M dataset, extracting features through Global Average Pooling (GAP)

Computer VisionTransformers

CLIP model with ResNet50 architecture trained on the CC12M dataset, supporting zero-shot image classification tasks

Computer VisionSafetensors

CC BY-NC 4.0 is a widely used Creative Commons license that allows users to freely share, use, and adapt works on the condition that the original author is credited and the works are not used for commercial purposes.

Natural Language ProcessingTransformers

A text-to-image generation model based on Stable Diffusion 1.5 architecture, specifically trained using CC0-licensed images and other freely usable content.

Computer VisionDiffusersOther

This is a CLIP-based ViT-B/32 model trained using a Data Filtering Network (DFN) on datasets including CC12M, CC3M, and Shutterstock 15M, suitable for zero-shot image classification tasks.

MultimodalTransformers

A video action recognition model based on the ViViT architecture, fine-tuned specifically for CCTV surveillance scenarios, excelling in action recognition tasks.

Computer VisionTransformers

ElanMT is a resource-efficient English-Japanese translation model developed by the ELAN MITSUA project/Abstract Engine. Despite relatively limited training resources, this model achieves performance comparable to current open-source translation models through back-translation and a newly constructed CC0 corpus. The entire training process uses only openly licensed data and Wikipedia back-translation data to ensure data purity.

Natural Language ProcessingTransformersMultiple Languages

A machine learning model specifically designed for real-time CCTV road traffic monitoring, used for road traffic flow estimation and metrics statistics

Computer VisionTransformersEnglish

OpenCLIP model based on ResNet50 architecture and trained on the CC12M dataset, supporting zero-shot image classification tasks.

Computer VisionSafetensors

The Creative Commons Attribution 4.0 International License (CC BY 4.0) is a license that allows users to freely share, use, and adapt works, provided that the original author is credited.

Natural Language ProcessingTransformers

MetaCLIP is a large-scale vision-language model trained on 2.5 billion data points from CommonCrawl (CC), revealing CLIP's data filtering methodology

MultimodalTransformers