okuvshynov/slowllama
A fine-tuning tool for Llama2 and CodeLlama models, including 70B/35B variants, on MacBook Air or consumer GPUs without quantization.
Not currently ranked — collecting fresh signals.
star history
slowllama enables fine-tuning of large language models on memory-constrained devices by offloading model components to SSD or main memory during both forward and backward passes. It uses LoRA (Low-Rank Adaptation) to limit parameter updates to a smaller set of weights, making training feasible on limited hardware. The project supports Llama2 and CodeLlama variants up to 70B parameters on Apple M1/M2 devices and consumer NVIDIA GPUs.
Frequently asked
- What is okuvshynov/slowllama?
- A fine-tuning tool for Llama2 and CodeLlama models, including 70B/35B variants, on MacBook Air or consumer GPUs without quantization.
- Is slowllama open source?
- Yes — okuvshynov/slowllama is open source, released under the MIT license.
- What language is slowllama written in?
- okuvshynov/slowllama is primarily written in Python.
- How popular is slowllama?
- okuvshynov/slowllama has 449 stars on GitHub.
- Where can I find slowllama?
- okuvshynov/slowllama is on GitHub at https://github.com/okuvshynov/slowllama.