Issues

• Reshaping a zero-size tensor panics with index-out-of-bounds in cubecl-zspace instead of a clean shape error

  tracel-ai/burn#6217 · Filter by author lileeei opened on Oct 25, 2026

• [Metal] bf16 Conv2d produces inf/NaN when weights are applied via burn-store load_from

  tracel-ai/burn#6216 · Filter by author lileeei opened on Oct 25, 2026

• cubecl: duplicate-safe scatter_nd reductions (bool without atomics)

  tracel-ai/burn#6213 · Filter by author HarpeLm opened on Oct 24, 2026

• Softmax Collapse: log_softmax, softmax and CrossEntropyLoss zero the target gradient at high confidence

  bug
  tracel-ai/burn#6212 · Filter by author antimora opened on Oct 24, 2026
  1

• Tensor::expand target Shape rank is not validated against output rank D2

  tracel-ai/burn#6210 · Filter by author onenewcode opened on Oct 24, 2026

• tensor: add dilation to AvgPoolOptions

  tracel-ai/burn#6148 · Filter by author antimora opened on Oct 20, 2026

• tensor: ONNX Resize coordinate transformation and rounding modes for interpolate

  tracel-ai/burn#6147 · Filter by author antimora opened on Oct 20, 2026
  1

• cubecl: work-efficient parallel scan for cumulative ops

  tracel-ai/burn#6122 · Filter by author saiyambharara opened on Oct 18, 2026

• Undefined behavior in pack_i8s_to_u32s on little-endian targets

  tracel-ai/burn#6117 · Filter by author onenewcode opened on Oct 17, 2026

• Lora with rank = 0 is not rejected and silently produces NaN weights

  tracel-ai/burn#6108 · Filter by author Kostyanikov opened on Oct 16, 2026

• JAX-like function transformations for differentiable programming in Burn

  feature
  tracel-ai/burn#6090 · Filter by author jpssrocha opened on Oct 16, 2026
  2

• train: add regression metrics (MSE, MAE, R²)

  tracel-ai/burn#6034 · Filter by author AlixANNERAUD opened on Oct 10, 2026

• int from_data with i64 data silently uses the device's default int dtype

  tracel-ai/burn#6006 · Filter by author altunenes opened on Oct 8, 2026
  6

• docs.rs fails to build burn-cuda, burn-rocm and burn-cpu (tracel-llvm-bundler)

  documentationinfrastructure
  tracel-ai/burn#5974 · Filter by author antimora opened on Oct 4, 2026

• [WGPU/Metal] Inference outputs wrong/saturated values; NdArray is correct

  bug
  tracel-ai/burn#5962 · Filter by author knarf-f opened on Oct 3, 2026
  7

• train: make DDP return Result instead of panicking

  refactor
  tracel-ai/burn#5959 · Filter by author Charles23R opened on Oct 2, 2026

• Apparent system memory leak on simple training task using WGPU backend

  bugwgpu
  tracel-ai/burn#5891 · Filter by author joshhansen opened on Sep 28, 2026 · patch-0.22.1
  19

• Add linalg::solve for square linear systems

  feature
  tracel-ai/burn#5855 · Filter by author Mikyx-1 opened on Sep 26, 2026

• cpu: segfault in depthwise conv backward

  blockerbug
  tracel-ai/burn#5802 · Filter by author Charles23R opened on Sep 23, 2026
  1

  

• tensor: scatter with smaller index shapes, bool assign

  enhancementonnx
  tracel-ai/burn#5784 · Filter by author antimora opened on Sep 23, 2026
  1

• cuda: graph-captured optimizers replay stale host scalars (learning rate)

  enhancement
  tracel-ai/burn#5779 · Filter by author swfsql opened on Sep 22, 2026
  2

• cuda: host time per step grows with the number of distinct shapes

  bugperformance
  tracel-ai/burn#5751 · Filter by author swfsql opened on Sep 20, 2026
  6

• train: automatic mixed precision (BF16 autocast, FP16 grad scaling)

  feature
  tracel-ai/burn#5696 · Filter by author Mikyx-1 opened on Sep 16, 2026

• train: lifecycle callbacks (per epoch, per batch)

  feature
  tracel-ai/burn#5689 · Filter by author fuzing opened on Sep 15, 2026
  2

• flex: PyTorch parity for half-precision scalar ops and comparisons

  enhancementflex
  tracel-ai/burn#5615 · Filter by author antimora opened on Sep 8, 2026
  1