Updated Dockerfile for DGX Spark #133
Conversation
These tweaks seem to work with DGX Spark
Summary of ChangesHello @sameersegal, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request updates the Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
The Dockerfile updates primarily focus on adapting the environment for DGX Spark, as indicated by the PR description. A minor path normalization for CUDA_HOME was applied. The most significant change involves modifying the pip install commands, specifically introducing the --no-deps flag for core libraries. While this might be intended for precise dependency control, it introduces a potential risk of missing runtime dependencies if not carefully managed, which could lead to unexpected failures.
|
|
||
| # Install unsloth and other dependencies | ||
| RUN pip install unsloth unsloth_zoo bitsandbytes==0.48.0 transformers==4.56.2 trl==0.22.2 | ||
| RUN pip install --no-deps bitsandbytes==0.48.0 transformers==4.56.2 trl==0.22.2 |
There was a problem hiding this comment.
The addition of --no-deps to the pip install command for bitsandbytes, transformers, and trl means these packages will be installed without their declared dependencies. This can lead to runtime errors if their required dependencies are not already satisfied by the base image or explicitly installed in a previous step. It's generally safer to let pip manage dependencies unless there's a very specific reason to override this behavior, and if so, those dependencies should be explicitly handled. Consider removing --no-deps for a more robust installation, or explicitly list and install all necessary dependencies for these packages.
RUN pip install bitsandbytes==0.48.0 transformers==4.56.2 trl==0.22.2
|
Oh thank you so much! |
2 participants
These tweaks seem to work with DGX Spark. I am able to run the GPT-OSS-20B 2048 notebook.
Solves issue #132