All issues

Search results

🐛 [Bug] arange passes a rank-1 tensor to LINSPACE's rank-0 alpha input; the API error is logged, not raised

Status: Open.

#4625 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] engine meta kernel reports the pre-truncation output dtype (float64) while the engine returns float32

Status: Open.

#4624 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] truncate_double inserts a float64 restoring cast but leaves the engine node's metadata claiming float64

Status: Open.

#4623 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] engine output symbols are reused verbatim across traces, silently asserting a bogus relation

Status: Open.

#4622 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] index_put with a None index materialises one index per element, failing on a runtime-known extent

Status: Open.

#4621 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] inlined execute_engine gets a SymInt where its schema declares Tensor[], breaking AOTI serialization

Status: Open.

#4620 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] output-dimension symbols recreated from engine metadata lose their size-like marking

Status: Open.

#4619 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] engine meta kernel has no ShapeEnv when inputs are statically shaped but outputs are symbolic

Status: Open.

#4618 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] engine meta kernel takes the output device from inputs[0], failing when the first input is a scalar

Status: Open.

#4617 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] engine meta kernel reads .shape on scalar inputs despite recording is_scalar: True

Status: Open.

#4616 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] partitioning ignores device placement, so an engine absorbs host work and returns it on the GPU

Status: Open.

#4615 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026

🐛 [Bug] a SymInt engine output is returned as a 0-d tensor and never converted back for scalar consumers

Status: Open.

#4614 In pytorch/TensorRT;

· SrivastavaKshitij opened on Aug 27, 2026