Lorenzo Sani

Codebase for the Photon project

Supercharge Your Model Training

No description.

The fastest local AI engine for Apple Silicon. 4.2x faster than Ollama, 0.08s cached TTFT, 100% tool calling. 17 tool parsers, prompt cache, reasoning separation, cloud routing. Drop-in OpenAI replacement. Works with Claude Code, Cursor, Aider.

Run LLMs with MLX

MLX-VLM is a package for inference and fine-tuning of Vision Language Models (VLMs) on your Mac using MLX.

Simulation framework for accelerating research in Private Federated Learning

LLM inference in C/C++

🤗 ml-intern: an open-source ML engineer that reads papers, trains models, and ships ML models

AI agents running research on single-GPU nanochat training automatically

The open source coding agent.

User-friendly AI Interface (Supports Ollama, OpenAI API, ...)

Better Harness Tools, not merely storing the archive of leaked Claude Code but also make real things done. Now rewriting in Rust.

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. All original source code is the property of Anthropic.

AllenAI's post-training codebase

PyTorch building blocks for the OLMo ecosystem

Reproducible, flexible LLM evaluations

A framework for few-shot evaluation of language models.

Flower: A Friendly Federated Learning Framework

LLM training code for MosaicML foundation models

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling

A high-throughput and memory-efficient inference and serving engine for LLMs

My personal webpage, built on the Github Pages template for academic personal websites, forked from mmistakes/minimal-mistakes

Fast and memory-efficient exact attention

A Data Streaming Library for Efficient Neural Network Training

No description.

Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends

No description.

No description.

No description.

FedScale is a scalable and extensible open-source federated learning (FL) platform.

Official implementation of Neurips 2020 "Sparse Weight Activation Training" paper.

An annotated implementation of the Transformer paper.

No description.

No description.

Toy project of federated learning

No description.

No description.