Add Troubleshooting Section to README #437
Open
+34
−3
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR enhances the README by adding a Troubleshooting section to help users resolve common issues they may encounter while using DeepSeek-V3.
New Section Added: Troubleshooting
The following issues and solutions are included:
Model weights not found: Instructions on downloading model weights from Hugging Face and placing them correctly.
CUDA errors during inference: Steps to ensure CUDA is set up correctly and PyTorch is configured for GPU use.
Slow inference performance: Recommendations for hardware optimization and using FP8/BF16 modes for faster inference.
Out of memory errors: Guidance on reducing batch sizes or leveraging model parallelism for multi-GPU setups.
Why This Change?
Users may face these common issues when running DeepSeek-V3. Adding a dedicated troubleshooting section improves usability and reduces potential support queries.
The troubleshooting tips are specific to the DeepSeek-V3 workflow and link to external resources for further guidance.
Please let me know if there are additional issues to include or if further edits are required.