espadrine
Follow
Stars
Moshi is a speech-text foundation model and full-duplex spoken dialogue framework. It uses Mimi, a state-of-the-art streaming neural audio codec.
Official inference repo for FLUX.1 models
Anole: An Open, Autoregressive and Native Multimodal Models for Interleaved Image-Text Generation
Haiku-Search is a high-performance fuzzy search library designed for web applications. It is built using Rust and compiled to WebAssembly
Inference and training library for high-quality TTS models.
Svart: A high-performance vector database in Rust. Named after the Swedish word for 'black,' Svart aims for sleek and efficient data indexing and retrieval.
A natural language interface for computers
The TinyLlama project is an open endeavor to pretrain a 1.1B Llama model on 3 trillion tokens.
Making Ascon cipher easier to play with
QLoRA: Efficient Finetuning of Quantized LLMs
🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading
Book repository "Analysis and Design of Elementary MOS Amplifier Stages"
A book teaching assembly language programming on the ARM 64 bit ISA. Along the way, good programming practices and insights into code development are offered which apply directly to higher level la…
Code repository for supporting the paper "Atlas Few-shot Learning with Retrieval Augmented Language Models",(https//arxiv.org/abs/2208.03299)
GLM-130B: An Open Bilingual Pre-Trained Model (ICLR 2023)
A repo for distributed training of language models with Reinforcement Learning via Human Feedback (RLHF)
Publication-ready NN-architecture schematics.
A latent text-to-image diffusion model
Self-hostable maps stack, powered by OpenStreetMap.
Deserialize JSON API formats into vanilla Ruby objects.
NVIDIA Linux open GPU kernel module source
Compilation des sondages produits à l'occasion des élections régionales et présidentielle en France, ainsi que d'autres jeux de données concernant les élections.