Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Conversation

@Ruturaj4
Copy link
Contributor

@Ruturaj4 Ruturaj4 commented Jun 5, 2025

This PR adds a new GitHub Actions workflow that:

Builds JAX with ROCm support inside a Docker container.
Runs training for the following MaxText models:

  • llama2_7b
  • gemma_2b
  • gpt3_6b
  • mixtral_8x1b

Captures stdout logs for each model and extracts per-step timing

Ignores step 0 (warmup) when computing metrics

Computes median_step_time per model and saves it to summary.json

Uploads logs and metrics as workflow artifacts

A Python analysis script (analyze_maxtext_logs.py) is added under jax/build/rocm/ to parse logs and generate the summary.

@Ruturaj4 Ruturaj4 force-pushed the bring_up_bench branch 7 times, most recently from ca207df to b2d9a83 Compare June 5, 2025 17:21
@JehandadKhan
Copy link
Collaborator

@Ruturaj4 Lint is failing .....

@Ruturaj4 Ruturaj4 force-pushed the bring_up_bench branch 18 times, most recently from 866b341 to cad81e1 Compare June 11, 2025 22:33
@Ruturaj4 Ruturaj4 force-pushed the bring_up_bench branch 4 times, most recently from 82bc9c0 to e69a17e Compare June 18, 2025 18:30
Copy link
Contributor

@psanal35 psanal35 left a comment

Choose a reason for hiding this comment

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

Looks good to me.

@Ruturaj4 Ruturaj4 merged commit ed6655c into master Jun 18, 2025
8 checks passed
@Ruturaj4 Ruturaj4 deleted the bring_up_bench branch June 18, 2025 21:56
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.

5 participants