Issues

• SGD applies weight decay to bias parameters as well

  enhancement
  Terminay/leanpass#66 · Filter by author github-actions opened on Sep 28, 2026

• Unnecessary Tensor allocation in Dropout forward pass

  performance
  Terminay/leanpass#65 · Filter by author github-actions opened on Sep 28, 2026

• Tensor.relu stores previous nodes as a set instead of a tuple

  bug
  Terminay/leanpass#63 · Filter by author github-actions opened on Sep 28, 2026

• Tensor.log silently clips non‑positive inputs

  enhancement
  Terminay/leanpass#64 · Filter by author github-actions opened on Sep 28, 2026

• Optimizer classes lack clear documentation of expected Tensor interface

  docs
  Terminay/leanpass#60 · Filter by author github-actions opened on Sep 21, 2026

• Division by zero raises exception even when NumPy could handle it

  enhancement
  Terminay/leanpass#59 · Filter by author github-actions opened on Sep 21, 2026

• Tensor.backward accumulates gradients across calls

  bug
  Terminay/leanpass#46 · Filter by author github-actions opened on Aug 31, 2026

• register_op silently overwrites existing ops

  bug
  Terminay/leanpass#44 · Filter by author github-actions opened on Aug 31, 2026

• Dropout scaling can overflow for p values close to 1

  bug
  Terminay/leanpass#38 · Filter by author github-actions opened on Aug 24, 2026

• Inconsistent representation of parent nodes in unary ops

  bug
  Terminay/leanpass#39 · Filter by author github-actions opened on Aug 24, 2026

• Numerical instability in log‑based loss functions

  enhancement
  Terminay/leanpass#36 · Filter by author github-actions opened on Aug 24, 2026

• Division by zero not guarded in Tensor.__truediv__ backward pass

  bug
  Terminay/leanpass#37 · Filter by author github-actions opened on Aug 24, 2026

• Linear layer does not validate in_features and out_features arguments

  bug
  Terminay/leanpass#34 · Filter by author github-actions opened on Aug 17, 2026

• Numerical instability in loss functions (cross_entropy_loss and binary_cross_entropy_loss)

  enhancement
  Terminay/leanpass#33 · Filter by author github-actions opened on Aug 17, 2026

• Tensor.log does not guard against non‑positive inputs

  bug
  Terminay/leanpass#24 · Filter by author github-actions opened on Aug 10, 2026

• Tensor.mean backward fails when axis is a tuple

  bug
  Terminay/leanpass#25 · Filter by author github-actions opened on Aug 10, 2026

• SGD.zero_grad overwrites gradients for non‑trainable parameters

  bug
  Terminay/leanpass#23 · Filter by author github-actions opened on Aug 10, 2026

• Adam.step bias‑correction may underflow for large step counts

  help wanted
  Terminay/leanpass#11 · Filter by author github-actions opened on Aug 8, 2026

• Duplicate and incomplete definitions of _graph_nodes and grad_check

  good first issue
  Terminay/leanpass#8 · Filter by author github-actions opened on Aug 8, 2026

• Docs say sum() and mean() are full-reduction only

  documentationgood first issue
  Terminay/leanpass#5 · Filter by author Terminay opened on Jul 30, 2026
  1

• tensor.md missing methods

  documentationgood first issue
  Terminay/leanpass#4 · Filter by author Terminay opened on Jul 30, 2026
  1