wx44wx/icassp2027-experiment
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Save periodic LoRA weight checkpoints

Browse Source
This commit is contained in:
Xin Wang
2026-08-24 15:31:44 +08:00
parent 7acab44ea2
commit 7fdda297d3
2 changed files with 81 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
README.md
View File

@@ -1154,6 +1154,16 @@ it be evaluated on official test `Sessions_1--3`. A second controlled run may
omit `--freeze-head` and use `--head-learning-rate 1e-5`; all other settings
must remain identical.
For a continuous five-epoch run, set
`TOTAL_STEPS=$((5 * MAX_STEPS))` and pass
`--max-steps "${TOTAL_STEPS}" --save-every-steps "${MAX_STEPS}"`. The trainer
keeps one AdamW/cosine-scheduler trajectory and writes evaluation-ready
`adapter/` plus `turn_head.pt` pairs under `checkpoint-step-MAX_STEPS`,
`checkpoint-step-2*MAX_STEPS`, through the fifth epoch. The top-level adapter
and head duplicate the final fifth checkpoint. Evaluate all five intermediate
checkpoints only on held-out `Session_12`, select one, then run that checkpoint
once on `Sessions_1--3`.
### Frozen-Thinker sparse head-only control
To determine whether the regression comes from transition-sparse sampling or