Skip to content

fix: parse training loss from updated Caffe log format#7128

Open
Chessing234 wants to merge 1 commit into
BVLC:masterfrom
Chessing234:fix/5422-parse-log-loss-extraction
Open

fix: parse training loss from updated Caffe log format#7128
Chessing234 wants to merge 1 commit into
BVLC:masterfrom
Chessing234:fix/5422-parse-log-loss-extraction

Conversation

@Chessing234

Copy link
Copy Markdown

Fixes #5422

Summary

  • Parse iteration numbers and loss values from lines containing loss =
  • Avoid splitting on the first = in timing lines like 36.0956s/100 iters), loss = 0.26044

Test plan

  • Run tools/extra/parse_log.sh against a log containing timing info in iteration lines
  • Confirm .train output puts numeric loss values in the TrainingLoss column

Made with Cursor

Extract loss values after the loss = marker so parse_log.sh works with
iteration timing lines that contain additional equals signs.

Fixes BVLC#5422

Co-authored-by: Cursor <cursoragent@cursor.com>
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.

parse_log.sh and plot_training_log.py broken by log format

1 participant