Projects sharing features with Meltingpot

This project is a language model evaluation framework and benchmarking tool designed to measure the accuracy and performance of models across diverse datasets. It provides a system for implementing model-based graders, running standardized tests for mathematical reasoning, coding, and factuality, and calculating quantified performance metrics such as precision, recall, F1 scores, and pass-at-k. The framework utilizes model-based grading and rubrics to validate response quality against expert-defined criteria. It includes a multi-model benchmarking loop and a model-agnostic API interface to co

SmolLM is a project dedicated to the development of small language models. It focuses on training and fine-tuning compact models that maintain high performance while utilizing fewer parameters. The project emphasizes efficient AI inference and on-device text generation, aiming to enable the deployment of lightweight models on edge devices with limited memory and processing power. It utilizes synthetic data generation to produce artificial datasets that improve the reasoning and training of these AI systems. The system supports a variety of optimization and training capabilities, including we

This project provides a structured methodology and framework for engineering, testing, and maintaining system instructions for large language models. It serves as a guide for designing clear, explicit prompts that ensure models follow complex tasks with consistent accuracy. The framework distinguishes itself by treating prompt development as a rigorous engineering process, emphasizing the use of version control to manage prompt templates and style guides. By organizing instructions into modular, reusable components, it facilitates long-term maintainability and reduces technical debt within de

This project is an LLM research orchestrator and autonomous AI agent framework designed to automate the scientific lifecycle. It functions as an end-to-end research pipeline and model training toolkit, managing everything from initial literature reviews and hypothesis testing to the final drafting of academic papers. The system is distinguished by its ability to convert unstructured academic PDFs into machine-executable knowledge layers, allowing agents to reproduce and extend research findings. It employs a two-loop orchestration architecture and a specialized research engineering skill libr

rllm is an asynchronous reinforcement learning framework for training language agents. It provides a unified pipeline that runs the same agent code for both evaluation and training, automatically capturing traces for gradient computation. The framework supports distributed reinforcement learning across multiple GPUs and nodes using pluggable backends, and executes agents in isolated sandboxes—either locally or in the cloud—for safe and scalable rollout collection. It trains agents built with LangGraph, SmolAgents, OpenAI Agents SDK, or custom frameworks without requiring core logic changes. T

MatchZoo is a deep learning framework designed for building, training, and evaluating neural networks that determine the relevance and similarity between pairs of textual inputs. It serves as a research platform for neural information retrieval, specifically supporting the development of models for document retrieval, question answering, and ranking tasks. The framework utilizes declarative architecture composition to define complex neural network structures. It includes automated hyper-parameter resolution to populate missing configuration parameters before model compilation and uses callbac

Osim-rl is a research environment designed for the development and evaluation of reinforcement learning agents within physics-based musculoskeletal simulations. It provides a standardized interface that maps physiological state observations to muscle excitation control signals, enabling the study of human movement and biomechanics through iterative policy optimization. The framework distinguishes itself by integrating high-fidelity musculoskeletal modeling with tools for scientific benchmarking and reproducible experimentation. It allows researchers to define custom reward functions and adjus

SlowFast is a PyTorch video understanding framework and spatiotemporal neural network library. It serves as a toolset for video action recognition, enabling the training and evaluation of models designed to classify complex activities and objects within video sequences. The framework is distinguished by its use of dual-pathway spatiotemporal sampling to capture both slow and fast motions. It supports self-supervised video learning for pre-training models on unlabeled data and employs multigrid spatiotemporal training to optimize learning across multiple spatial and temporal resolutions. The

This project is a technical curriculum and development guide focused on large language model prompt engineering, fine-tuning, and the creation of retrieval augmented generation applications. It serves as a comprehensive resource for developers to master crafting precise instructions and textual patterns to improve the quality and predictability of model outputs. The material covers the end-to-end workflow of adapting open-source models to specific datasets and integrating language models with vector databases to generate responses based on private information. It also provides a systematic ap

TCN is a deep learning sequence framework and library for building temporal convolutional networks. It provides a toolkit for implementing purely convolutional architectures to model sequential data as an alternative to recurrent neural networks. The project includes a sequence modeling benchmark suite designed to evaluate the accuracy and processing speed of architectures. This suite utilizes standardized tasks, including memory problems, digit classification, music, and language tasks, to quantify performance. The framework covers a range of structural components for sequence processing, s

Tinker Cookbook is an open-source framework for fine-tuning large language models, supporting supervised learning, reinforcement learning, and parameter-efficient techniques like LoRA adapters. It provides a complete pipeline for aligning models with human preferences through multi-stage RLHF workflows, from supervised fine-tuning through preference optimization to reinforcement learning. The framework distinguishes itself through recipe-based training orchestration, where fine-tuning workflows are defined as composable recipe files that chain data loading, model configuration, and training l

jetson-inference is a set of libraries and tools for executing optimized deep learning models on embedded GPU hardware. Its primary purpose is to enable real-time computer vision and AI inference at the edge with low latency and high throughput. The project distinguishes itself through high-performance streaming analytics and the ability to execute concurrent AI pipelines on auto-grade silicon. It provides specialized support for multi-sensor stream processing, utilizing zero-copy data transport to load camera frames directly into GPU memory. The codebase covers a broad surface of capabiliti

A replica of the AlphaZero methodology for deep reinforcement learning in Python

This repository provides supplementary material for our paper Constitutional AI: Harmlessness from AI Feedback.

⚡️A Blazing-Fast Python Library for Ranking Evaluation, Comparison, and Fusion 🐍

RewardBench: the first evaluation tool for reward models.

Welcome to drlzh.ai: a hands-on deep reinforcement learning course where you build the algorithms, not just read about them.

TensorFlow implementation of Deep Reinforcement Learning papers

Accelerated RL

|tests| |pypi| |docs| |License|

Opik is an observability and evaluation platform designed for generative AI applications and agentic workflows. It provides a centralized environment for tracing execution flows, managing prompt templates, and monitoring production performance, allowing teams to gain visibility into complex model interactions and tool usage without requiring manual application code changes. The platform distinguishes itself through its integrated approach to the AI development lifecycle, combining distributed trace instrumentation with automated evaluation frameworks. It supports model-as-a-judge scoring, syn

FinRL is a reinforcement learning framework designed for the development, training, and backtesting of automated trading strategies. It functions as a quantitative finance toolkit that integrates deep learning algorithms with financial market simulations to address complex portfolio management and asset allocation tasks. The platform provides an end-to-end pipeline for transforming raw market data into actionable trading models. The project distinguishes itself through a layered, modular architecture that separates data processing, environment simulation, and agent training. This design allow