OpenMOSS/CoLLiE
A toolkit for collaborative and efficient training of large language models, built on PyTorch and DeepSpeed.
Not currently ranked — collecting fresh signals.
star history
CoLLiE provides a complete toolbox for training large language models from scratch, enabling collaborative and efficient distributed training across GPUs and nodes. It integrates with DeepSpeed and PyTorch to handle parallelism, gradient accumulation, and optimization strategies for training LLMs at scale. The project supports various LLM architectures and provides benchmarking utilities for evaluating training efficiency.
Frequently asked
- What is OpenMOSS/CoLLiE?
- A toolkit for collaborative and efficient training of large language models, built on PyTorch and DeepSpeed.
- Is CoLLiE open source?
- Yes — OpenMOSS/CoLLiE is open source, released under the Apache-2.0 license.
- What language is CoLLiE written in?
- OpenMOSS/CoLLiE is primarily written in Python.
- How popular is CoLLiE?
- OpenMOSS/CoLLiE has 421 stars on GitHub.
- Where can I find CoLLiE?
- OpenMOSS/CoLLiE is on GitHub at https://github.com/OpenMOSS/CoLLiE.