Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix: inference fault tolerance #108

Merged
merged 23 commits into from
Nov 2, 2023
Merged

Conversation

ishaansehgal99
Copy link
Collaborator

@ishaansehgal99 ishaansehgal99 commented Oct 26, 2023

This PR introduces a couple changes:

  1. Minor fixes on the e2e preset build pipeline
  2. Updating image version of pytorch/cuda/nccl from 23.06 to 23.10
  3. Update a docker file path (/home/llama -> /llama and /home/falcon -> /falcon) to prevent conflicting dir paths (host-volume: /home) mounting over one another
  4. Add torch rdzv and headless service for inference fault tolerance
  5. Inference code increased error handling, resiliency, fault tolerance, timeouts

@ishaansehgal99 ishaansehgal99 changed the title fix: worker networking issue inference fix: inference fault tolerance Nov 2, 2023
Copy link
Collaborator

@Fei-Guo Fei-Guo left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Rename createHeadlessService to useHeadlessService in all cases.

presets/llama-2-chat/inference-api.py Outdated Show resolved Hide resolved
presets/llama-2-chat/inference-api.py Outdated Show resolved Hide resolved
presets/llama-2/inference-api.py Outdated Show resolved Hide resolved
@Fei-Guo Fei-Guo merged commit d0ba4d5 into main Nov 2, 2023
5 checks passed
@Fei-Guo Fei-Guo deleted the Ishaan/fix-worker-networking-inference branch November 2, 2023 21:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants