search

AlignmentResearch / tuned-lens

Tools for understanding how transformer predictions are built layer-by-layer
https://tuned-lens.readthedocs.io/en/latest/
MIT License
437 stars 47 forks source link

Improve logging durring training and evaluation #111

Closed levmckinney closed 1 year ago

levmckinney commented 1 year ago

Resolves #86

codecov[bot] commented 1 year ago

Codecov Report

Merging #111 (bac6d08) into main (9bf1f35) will increase coverage by 0.12%. The diff coverage is 79.06%.

Impacted file tree graph

@@            Coverage Diff             @@
##             main     #111      +/-   ##
==========================================
+ Coverage   80.95%   81.08%   +0.12%     
==========================================
  Files          32       32              
  Lines        2116     2141      +25     
==========================================
+ Hits         1713     1736      +23     
- Misses        403      405       +2
Files Changed Coverage Δ
tests/scripts/test_integration.py 100.00% <ø> (ø)
tuned_lens/nn/lenses.py 91.89% <50.00%> (+0.07%) :arrow_up:
tuned_lens/causal/subspaces.py 25.83% <66.66%> (+1.25%) :arrow_up:
tuned_lens/scripts/train_loop.py 78.24% <75.00%> (-0.06%) :arrow_down:
tuned_lens/scripts/ingredients.py 86.18% <81.81%> (+0.55%) :arrow_up:
tuned_lens/__main__.py 96.29% <90.00%> (-3.71%) :arrow_down:
tuned_lens/scripts/eval_loop.py 90.36% <100.00%> (+0.11%) :arrow_up: