All pull requests

New pull request

Open

(66)

Closed

(1,833)

Preserve source locations in visitor transforms
#2834mukeshr06 opened Sep 22, 2026
[DO NOT MERGE] Resolve CI: torchaudio, dynamo FutureWarning, torch 2.13 named tensors
#2832bhimrazy opened Sep 15, 2026
2
Drop unused torchaudio from devel requirements
#2831bhimrazy opened Sep 15, 2026
Ignore deprecated dynamo config FutureWarning from torch test utils
#2830bhimrazy opened Aug 13, 2026
Fix import thunder on PyTorch 2.13 (named tensor removal)
#2829bhimrazy opened Aug 13, 2026
Remove F405 from Ruff ignore list
#2828arubittu opened Aug 12, 2026
2
Support PyTorch 2.12 (fx codegen, autograd_function_apply)
#2825bhimrazy opened May 29, 2026
chore: drop Azure CI configurations
#2819deependujha opened Apr 4, 2026
Smoke test
#2814wujingyue opened Feb 2, 2026
Enable symbolic values and count recompiles in benchmark scripts
#2801shino16 opened Dec 13, 2025
Add TileGym executor for SDPA and RMSNorm
#2800crcrpar opened Dec 13, 2025
[WIP] See if we can make symbolic values default
#2795shino16 opened Dec 10, 2025
1
Implement setitem without copy_
#2792shino16 opened Dec 9, 2025
2
[benchmark_inference.py] Specify tp_size to StaticCache
#2784crcrpar opened Dec 3, 2025
2
Bump the gha-updates group across 1 directory with 3 updates
#2783dependabot[bot] opened Dec 1, 2025
Bump litgpt from 0.5.0 to 0.5.11
#2782dependabot[bot] opened Dec 1, 2025
Update coverage requirement from ~=7.9.1 to ~=7.12.0
#2781dependabot[bot] opened Dec 1, 2025
Bump transformers from 4.52.4 to 4.57.3
#2780dependabot[bot] opened Dec 1, 2025
Make in-place ops DCE-able after update_aliases
#2777shino16 opened Nov 27, 2025
38
Remove unnecessary repetition of update_aliases
#2772shino16 opened Nov 26, 2025
20
[benchmark_inference] torchao nvfp4
#2770crcrpar opened Nov 26, 2025
Use torch._logging._internal.trace_structured_artifact to save TraceCtxs when TORCH_TRACE is set
#2751crcrpar opened Nov 18, 2025
28
Add F.scaled_grouped_mm
#2747crcrpar opened Nov 18, 2025
7
Remove ChunkedHybridCache from benchmark_inference.py
#2733IvanYashchuk opened Nov 13, 2025
10
Use torch._grouped_mm in eager mode
#2721wujingyue opened Nov 6, 2025
4

Navigation Menu

All pull requests

Preserve source locations in visitor transforms

[DO NOT MERGE] Resolve CI: torchaudio, dynamo FutureWarning, torch 2.13 named tensors

Drop unused torchaudio from devel requirements

Ignore deprecated dynamo config FutureWarning from torch test utils

Fix `import thunder` on PyTorch 2.13 (named tensor removal)

Remove F405 from Ruff ignore list

Support PyTorch 2.12 (fx codegen, autograd_function_apply)

chore: drop Azure CI configurations

Smoke test

Enable symbolic values and count recompiles in benchmark scripts

Add TileGym executor for SDPA and RMSNorm

[WIP] See if we can make symbolic values default

Implement `setitem` without copy_

[benchmark_inference.py] Specify `tp_size` to `StaticCache`

Bump the gha-updates group across 1 directory with 3 updates

Bump litgpt from 0.5.0 to 0.5.11

Update coverage requirement from ~=7.9.1 to ~=7.12.0

Bump transformers from 4.52.4 to 4.57.3

Make in-place ops DCE-able after `update_aliases`

Remove unnecessary repetition of `update_aliases`

[benchmark_inference] torchao nvfp4

Use `torch._logging._internal.trace_structured_artifact` to save `TraceCtx`s when `TORCH_TRACE` is set

Add `F.scaled_grouped_mm`

Remove ChunkedHybridCache from benchmark_inference.py

Use torch._grouped_mm in eager mode

Pull requests