Bram Wasti

Remove std::cerr from CachingMemoryManager.cpp

1

This was printed every time any Flashlight program runs. Also, doesn't seem to be formatted correctly

Add a SpanDataset

3

Add a SpanDataset to simplify splitting training and validation data from merged initial datasets **Original Issue**: closes #976 ### Summary As explained in the issue, this feature simplifies splitting up...

### Feature Description Not all datasets can be easily split ahead of time, a SpanDataset would allow splitting up datasets after the fact by specifying an offset and bound for...

CUDA assumption in the ts_compile code

2

Hey folks, stumbled into a CUDA assumption (on my non-CUDA machine) Here's the fix for me, but it's obviously not very general ``` diff --git a/functorch/_src/compilers.py b/functorch/_src/compilers.py index 10fe42a..83002ac 100644...

Feature request: Add an example of PyTorch's canonical error checking in the CUDA example

### Bug Description As in title https://github.com/flashlight/flashlight/blob/5b4441994e7742ac08793204cccd74a73e9079c3/flashlight/fl/tensor/backend/af/ArrayFireReductions.cpp#L184-L192 (current workaround in shumai: https://github.com/facebookresearch/shumai/blob/main/shumai/cpp/binding_gen.inl#L454-L458)

Segfault on large number of process spawns

1

### Version 0.2.1 ### Platform Darwin bwasti-mbp 21.6.0 Darwin Kernel Version 21.6.0: Mon Aug 22 20:19:52 PDT 2022; root:xnu-8020.140.49~2/RELEASE_ARM64_T6000 arm64 ### What steps will reproduce the bug? ```javascript import {...

`WebSocket` (client) hang on repeated large buffer sends

5

### Version 0.2.1 ### Platform Darwin bwasti-mbp 21.6.0 Darwin Kernel Version 21.6.0: Mon Aug 22 20:19:52 PDT 2022; root:xnu-8020.140.49~2/RELEASE_ARM64_T6000 arm64 ### What steps will reproduce the bug? Using the README...

Enable smaller "period" increments

1

Yahoo's API supports minute-level timescales for historical quotes.

Detect the C++ ABI of LLVM and PyTorch libs, ensure they are compatible and then build torch_tvm accordingly.