Open-source alternatives to Chemcrow Public

AgentLaboratory is a multi-agent research system that automates the entire scientific experimentation process, from literature review through experiment execution to report generation, using a sequence of specialized AI agents. The system orchestrates a team of language-model-driven agents—a literature reviewer, experimental planner, executor, and report writer—to autonomously complete an end-to-end research workflow. The system distinguishes itself by saving progress at every checkpoint, enabling seamless recovery and continuation after interruptions or failures. Agents build on each other's

This project is a collection of specialized toolkits and an agent skill library designed to equip large language model agents with the capabilities to perform complex scientific research across biology, chemistry, medicine, and physics. It provides a structured framework of integration paths and tools that allow agents to execute multi-step research workflows. The system is distinguished by its domain-specific toolsets, including a bioinformatics toolkit for genomic and single-cell analysis, a cheminformatics toolset for drug-target binding and lead compound optimization, and a multi-omics an

AutoResearchClaw is an agentic system designed to automate the scientific research process. It functions as an autonomous research agent and workflow automator that manages the entire lifecycle of a project, from initial hypothesis generation and literature review to experimental execution and the production of LaTeX-formatted academic papers. The system distinguishes itself through a multi-agent research pipeline that utilizes structured debates for hypothesis refinement and peer review. It employs a branch-and-merge architecture to explore parallel research directions and integrates human-i

AI-Scientist is an autonomous research pipeline and framework for scientific discovery. It employs large language model agents to manage the full lifecycle of a scientific project, from initial hypothesis generation to the production of formal academic papers. The system operates through a repeating research loop that integrates automated experimental execution, data analysis, and literature novelty verification. It queries external academic databases to validate the originality of ideas and retrieve citations, then translates experimental findings into LaTeX manuscripts. To refine these outp

TinyLlama is a compact 1.1B parameter language model pretrained on a dataset of 3 trillion tokens. It is an edge AI model designed for high-performance text generation on memory-constrained devices. The project provides a distributed pretraining framework for training small language models across multiple GPUs and nodes. It also includes a finetuning toolkit for full-parameter weight adjustments to adapt the base model for chat and specific tasks. The system supports distributed large language model training and on-device text generation. Its architectural components include rotary positiona

This project is a scientific agent framework and workflow orchestrator designed to extend large language models with specialized tools for genomic, chemical, and biological research. It provides a system for planning research hypotheses and executing automated workflows by integrating scientific databases with dynamic code execution. The framework includes a cheminformatics modeling suite for predicting molecular bioactivity and performing virtual screening, alongside a bioinformatics analysis toolkit for processing genomic sequences and single-cell data. It also features an academic document

This project is a comprehensive framework for building, evaluating, and connecting autonomous agent systems. It provides a library of standardized architectural patterns for implementing complex agent workflows, including multi-agent orchestration, iterative reasoning, and memory management. By offering a unified interface for model providers, the framework allows for consistent agent execution across different artificial intelligence services. The framework distinguishes itself through a focus on rigorous benchmarking and deterministic control. It includes a suite of tools for evaluating age

OpenDeepSearch is an autonomous research platform and search orchestration engine that connects large language models with live web data. It functions as an agentic information retrieval tool designed to automate the gathering, structuring, and synthesis of web information to resolve complex queries. The system differentiates itself through iterative multi-hop querying, which allows the agent to execute sequences of dependent searches where previous results refine subsequent requests. It features a semantic reranking pipeline that uses embedding models to prioritize the most relevant content

Gemma is a family of open-weights large language models based on a decoder-only transformer architecture. These models are designed for text generation and multi-modal conversations, capable of processing and generating responses based on both textual and visual input sequences. The project provides a fine-tunable AI model that supports weight adjustment and low-rank adaptation to specialize performance for particular tasks. It includes support for quantized weights to reduce memory usage and increase inference speed on limited hardware. The capability surface covers multi-modal AI integrati

ChemBench

BELLE is a specialized implementation of Chinese conversational large language models, encompassing a full instruction tuning framework. It provides a pipeline for training, evaluating, and deploying models optimized for natural language understanding and dialogue tasks in the Chinese language. The project is distinguished by its integrated approach to model refinement, combining the curation of multi-million entry instruction datasets with a distributed training pipeline. This pipeline supports both full fine-tuning and low-rank adaptation to optimize conversational performance. The system

Code for fintune ChatGLM-6b using low-rank adaptation (LoRA)

A minimalist multi-agent framework for rubost automation of scientific analysis workflows, such as gene expression analysis.

FastChat is a training and serving platform for large language models that provides an integrated toolkit for fine-tuning, hosting, and benchmarking chatbots. It functions as an inference server capable of hosting multiple models and exposing them via a standardized API for chat applications. The platform distinguishes itself through a distributed model controller that manages worker nodes and routes requests across a hardware-agnostic inference layer supporting various accelerators. It includes a dedicated evaluation framework for assessing model quality using automated judges, multi-turn di

Develop and deploy smart contracts using natural language.

UFO is a multi-device task orchestrator and LLM agent orchestration framework designed to decompose natural language requests into executable task graphs. It functions as a cross-platform UI automation tool capable of performing interactions on Windows and mobile devices while routing tasks to distributed agents based on their hardware and software capabilities. The system is distinguished by its RAG-enhanced agent architecture, which integrates external documentation and previous execution traces to improve decision-making. It employs a hybrid UI detection approach that combines computer vis

Visual-ChatGPT is a visual orchestration framework and multimodal AI pipeline designed to coordinate large language models with visual foundation models. It functions as an integration layer that enables the exchange of text and images between different AI models to automate image analysis and editing tasks without requiring additional model training. The system differentiates itself through model-chain orchestration and prompt-based task dispatching, allowing natural language instructions to trigger specific vision models or tools. It utilizes coordinate-based region mapping and iterative ma

Kimi-K2 is a conversational AI engine and reasoning framework designed for text generation, advanced problem solving, and coding tasks. It functions as a tool-augmented language model capable of producing human-like chat responses through a compatible model interface. The system utilizes a reasoning-optimized architecture that separates standard conversational flow from deep logical processing. This allows the model to execute autonomous tasks by invoking external functions and calling APIs to retrieve real-time data. The project supports structured JSON output parsing for function-call inte

GPT4All is a cross-platform runtime environment designed to execute large language models directly on local consumer hardware. By leveraging an optimized C++ inference backend, it enables private, offline AI interactions without requiring an internet connection or external cloud services. The project provides a comprehensive ecosystem for managing the entire model lifecycle, including discovery, downloading, and configuration of local weights. What distinguishes the platform is its integrated retrieval-augmented generation engine, which allows users to index local documents into semantic vect

Code for the paper "Evaluating Large Language Models Trained on Code"

Tool Learning for Big Models, Open-Source Solutions of ChatGPT-Plugins

ToolBench is an open platform for training, serving, and evaluating large language models that retrieve and call real-world APIs to complete user instructions. It provides an API-aware inference engine that selects relevant tools from a large corpus and generates sequences of tool calls to produce final answers, along with a custom API registration system that lets users add their own REST endpoints for the model to discover and invoke. The platform includes a complete instruction-tuning pipeline for training models on curated tool-use data, a multi-tool execution engine that coordinates sequ

MOSS is a conversational AI platform, fine-tuning toolkit, and quantized model runtime. It provides a framework for deploying large language models capable of multi-turn dialogue, general-purpose response generation, and following complex instructions. The system functions as a tool-augmented framework that extends model knowledge through external plugins and tool-call loops. This allows the model to execute tasks via search engines and calculators to augment responses with external data. The project covers model training through supervised conversational fine-tuning and optimizes deployment

This is the official code repository for the paper LlaSMol: Advancing Large Language Models for Chemistry with a Large-Scale, Comprehensive, High-Quality Instruction Tuning Dataset.

ERNIE is a development toolkit for training, fine-tuning, and deploying large language models built on the PaddlePaddle deep learning platform. It provides a comprehensive suite of core components, including an inference server for vision and language models, a training and fine-tuning toolkit, and a framework for building retrieval-augmented generation systems using private knowledge bases. The project features multimodal AI models capable of reasoning across text, images, and video to perform complex visual understanding and information extraction. It distinguishes itself through specialize

)](https://rye.astral.sh/)