All issues

Search results

A subscript's shape is pinned for the dataflow analysis but not for a generator reading it as an argument

Status: Open.

#825 In wala/ML;

· khatchad opened on Aug 5, 2026

tf.gather returns ⊤ when the indices' shape is unknown, discarding the table's known trailing axes

Status: Open.

#823 In wala/ML;

· khatchad opened on Aug 5, 2026

A layer's explicit dtype= policy is not read, so the compute-dtype cast assumes the default

Status: Open.

#822 In wala/ML;

· khatchad opened on Aug 5, 2026

tf.keras.Sequential is unmodeled: no summary, so Sequential-built models carry no framework-typed instance

Status: Open.

#817 In wala/ML;

· khatchad opened on Aug 4, 2026

np.zeros with a complex dtype argument types as float64

Status: Open.

#816 In wala/ML;

· khatchad opened on Aug 1, 2026

Batch axis stays wildcarded where the pipeline drops the remainder

Status: Open.

#812 In wala/ML;

· khatchad opened on Jul 30, 2026

Layer widths from a configuration constant stay wildcarded across a module

Status: Open.

#811 In wala/ML;

· khatchad opened on Jul 30, 2026

Token-mask helpers infer unknown rank where the body requires rank 2

Status: Open.

#810 In wala/ML;

· khatchad opened on Jul 29, 2026

Model dimensions from constructor arguments stay wildcarded on layer inputs and logits

Status: Open.

#809 In wala/ML;

· khatchad opened on Jul 29, 2026

Source corpus-fixture expectations from runtime traces

workstream:precisionAriadne precision (corpus-fixture audit)Ariadne precision (corpus-fixture audit)

Status: Open.

#808 In wala/ML;

· khatchad opened on Jul 29, 2026

Binary-operator results mint no InstanceKey, cascading empty points-to sets

Status: Open.

#805 In wala/ML;

· khatchad opened on Jul 29, 2026

Scalar tf.Variable operand defeats the one-sided elementwise broadcast

Status: Open.

#804 In wala/ML;

· khatchad opened on Jul 29, 2026