The pinned upstream neural_net_mod module is compiled with an audited, generated fixed-weight model: two inputs, two ReLU hidden units and one output in REAL32. Operator generation uses the author Model/Dense/Activation API and NumPy 2.2.6; neither TensorFlow, PyTorch nor Python inference runs in visitor jobs. The generator sorts the imported layer routine names for reproducibility; network operations are unchanged. Four inputs exercise positive and clipped hidden units and predictions are checked against independently evaluated weights and biases, including negative and positive outputs and finiteness. K covers inference as approximation; no least-squares fitting or training classification is inferred. The sample network is illustrative, not a trained scientific model. Other network layers, external model conversion and training are not verified. BSD-3-Clause and Crown copyright notices are retained.
Package dependencies
Resolved GitHub repositories used by the verified playground build, combining direct and indirect dependencies. This list does not describe internal module dependencies or native system libraries.
Required by this build
No additional GitHub package dependencies were recorded for this verified build.
Build record verified 2026-10-30.
Used by (0 selected packages)
No other selected build records this repository as a dependency. This is not a measure of usage outside the catalog.
Reviewed procedures
These mappings have pinned source evidence. A checked package example may exercise only some of them; source review is not a numerical benchmark.
The pinned upstream neural_net_mod module is compiled with an audited, generated fixed-weight model: two inputs, two ReLU hidden units and one output in REAL32. Operator generation uses the author Model/Dense/Activation API and NumPy 2.2.6; neither TensorFlow, PyTorch nor Python inference runs in visitor jobs. The generator sorts the imported layer routine names for reproducibility; network operations are unchanged. Four inputs exercise positive and clipped hidden units and predictions are checked against independently evaluated weights and biases, including negative and positive outputs and finiteness. K covers inference as approximation; no least-squares fitting or training classification is inferred. The sample network is illustrative, not a trained scientific model. Other network layers, external model conversion and training are not verified. BSD-3-Clause and Crown copyright notices are retained. The dedicated GAMS driver additionally checks generic dense/activation pipelines with 1/2/5 channel rows and five activations against 35,200 scalar references, and the existing generated model on an 81-point grid covering all four ReLU regions. Extreme-logit softmax and external model conversion remain unverified.