AIwaves-cn/agents is an open-source Python framework for autonomous language agents, focusing on symbolic learning and self-evolving agents. It provides installation via git and supports local development.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Agents is an open-source framework for data-centric, self-evolving autonomous language agents. It implements loss function, back-propagation, and weight optimizer in the context of agent training using language-based loss, gradients, and weights. The project describes a workflow where an agent's execution (forward pass) is stored as a trajectory with input, output, prompts, and tool usage, then a language loss is computed and back-propagated to generate language gradients that update symbolic components and the computational graph. It supports multi-agent considerations by treating nodes as different agents or enabling multiple agents to take actions within a node.
How it works
The README outlines a workflow where a forward pass is followed by storing trajectory data, applying a prompt-based loss to obtain a language loss, back-propagating this loss from last to first node to obtain language gradients, and updating symbolic components and the graph accordingly. Loss, back-propagation, and weight update are designed for the context of language-based components within each node.
Getting started
Installation from git repo branch:
pip install git+https://github.com/aiwaves-cn/agents@master
Installation for local development:
git clone -b master https://github.com/aiwaves-cn/agents
cd agents
pip install -e .
Recent releases
- latest release line in README shows version banner v2.0.0, but there are no explicit release entries listed under RELEASES (latest 0): none.
Traction
Stars: 5955 Forks: 481 Open issues: 46
Caveats
License: Apache-2.0 Language: Python Created: 2023-07-18 Last push: 2024-09-26 Topics: autonomous-agents, language-model, llm