1
83%
main: 83%

Ran 11 Jul 2026 11:44PM UTC

Files 33

Run time 0s

Badge

Embed ▾

Committed 11 Jul 2026 11:35PM UTC coverage: 81.832%. Remained the same

Job # #2484.1

Build Type

push

github

Committed by

web-flow

Commit Message

Pin the side-effect precondition's precision across the strategy.run boundary (#771)

Diagnoses #768: the MusicTransformer `__dist_train_step` declines trace to genuine Python instance-field writes, `self.loss_value` and `self.grad`, present in both `__train_step` variants, not to TensorFlow-operation mutations. The fixture distills the corpus shape with a faithful trainer and an all-local control: the trainer and its `strategy.run` forwarder are charged transitively, while the control, whose reached path mutates only `tf.Variable` state via `apply_gradients`, is not charged at all. The precondition is therefore precise on this shape, refuting the false-positive hypothesis, and the declines are correct by the precondition's own definition.

Closes #768.


Claude-Session: https://claude.ai/code/session_01M4XLK81HCB6Vp1WQiRPafR

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

Coverage Stats

2135 of 2609 relevant lines covered (81.83%)

0.82 hits per line

ponder-lab / Hybridize-Functions-Refactoring / #2484 / 1
83%
main: 83%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job #2484.1

ponder-lab / Hybridize-Functions-Refactoring / #2484 / 1 83% main: 83%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job #2484.1

ponder-lab / Hybridize-Functions-Refactoring / #2484 / 1
83%
main: 83%

README BADGES
x