Directory
Published directory entries matching your search.
A personal website of Chris Albon. While browsing, the all-MiniLM-L6-v2 model runs on the visitor's computer and turns passages from congressional hearings into 384-number embeddings, which are compared to connect similar passages. Completed embeddings are saved to a shared collection, and visitors can inspect the numbers, download embeddings or pause the model.
Patent Architect AI v2 is a Hugging Face Space that takes a description of an invention and returns a patent strategy including prior art analysis, technical summary, figures, and claims. Users may optionally agree to share their data to help improve the tool. The Space is sleeping due to inactivity.
Retrieval-based-Voice-Conversion-WebUI is a project for training voice conversion models from voice data, with an easy-to-use WebUI. It includes a pymss/MSST model to separate vocals and instruments and a high-pitch voice extraction algorithm to prevent a muted sound problem. Dependencies differ by hardware, including CPU, AMD, Intel and NVIDIA GPU setups.
MMAudio is a CVPR 2025 project for video-to-audio synthesis that generates synchronized audio from video and/or text inputs. It uses multimodal joint training and a synchronization module to align generated audio with video frames. A pretrained large_44k_v2 model is used by default, with inference around 6GB of GPU memory in 16-bit mode.
Accelerates transcription with the combination of OpenAI's Whisper Large v2, HF Transformers, Optimum, and flash attention.
RedPajama-Data-v2 is a 30 trillion token open dataset from 84 CommonCrawl dumps across five languages, with over 40 precomputed quality annotations. Data processing scripts are open source on GitHub and all data are available on HuggingFace. The dataset is intended for further filtering and weighting of data, and the release notes describe it as an open dataset for training large language models.
P-tuning v2 is an optimized deep prompt tuning strategy that applies continuous prompts to every layer input of a pretrained transformer. It is described as comparable to fine-tuning across scales and tasks.
Nexusflow/NexusRaven-V2-13B is a 13-billion parameter language model available on Hugging Face. It is designed for zero-shot function calling and text generation, aiming to advance open-source artificial intelligence. The model can be used with libraries such as Transformers and provides instructions for deployment via inference providers, notebooks, and local applications.
Free production platform for text-to-image generation using Nano Banana V2 model.
