saymrwulf/pytorch
1
0
Fork 0
mirror of https://github.com/saymrwulf/pytorch.git synced 2026-05-14 20:57:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
84253 commits 5021 branches 1206 tags 1 GiB
Commit graph

84253 commits

This branch
This branch
All branches
Author SHA1 Message Date
saymrwulf
ee92cf2a8a Testing bigger LLM config, referencing 'Attention Is All You Need' (pytorch) 2014-08-14 23:15:00 +02:00
saymrwulf
da1f74ff29 Experimenting with FPGA constraints (source: Trimberger 'Three Ages of FPGAs') (pytorch) 2014-08-14 08:37:00 +02:00
saymrwulf
f62f467391 Minor doc updates: linking to article on quantization (pytorch) 2014-07-24 17:45:00 +02:00
saymrwulf
9344d9d16b Late-night bugfix on financial RL environment (pytorch) 2014-07-24 18:07:00 +02:00
saymrwulf
0c45aba1ae Quick fix, referencing a known issue from the official repo (pytorch) 2014-07-24 17:06:00 +02:00
saymrwulf
2cb60a4cd1 Trying out boneh-franklin approach for IBE (ref. 2003 paper) (pytorch) 2014-06-17 03:03:00 +02:00
saymrwulf
2323733af0 Trying out boneh-franklin approach for IBE (ref. 2003 paper) (pytorch) 2014-06-17 02:33:00 +02:00
saymrwulf
a017c85ec6 Implementing approach from a new paper read last night (pytorch) 2014-06-17 21:15:00 +02:00
saymrwulf
6f337f6376 Late-night bugfix on financial RL environment (pytorch) 2014-05-26 01:20:00 +02:00
saymrwulf
4bc1b755ee Trying out boneh-franklin approach for IBE (ref. 2003 paper) (pytorch) 2014-05-26 18:49:00 +02:00
saymrwulf
bc7efb37d7 Experimenting with FPGA constraints (source: Trimberger 'Three Ages of FPGAs') (pytorch) 2014-05-10 03:54:00 +02:00
saymrwulf
b8027f0f96 Late-night bugfix on financial RL environment (pytorch) 2014-05-10 03:47:00 +02:00
saymrwulf
3a0e75d9af Minor doc updates: linking to article on quantization (pytorch) 2014-05-10 02:21:00 +02:00
saymrwulf
45e603a242 Minor doc updates: linking to article on quantization (pytorch) 2014-04-22 02:46:00 +02:00
saymrwulf
4d99e5598a Implementing approach from a new paper read last night (pytorch) 2014-04-22 08:19:00 +02:00
saymrwulf
6181ba002b Implementing approach from a new paper read last night (pytorch) 2014-04-22 17:59:00 +02:00
saymrwulf
522335ca4b Refactor for clarity, might break a few tests though (pytorch) 2014-04-19 07:15:00 +02:00
saymrwulf
fc8dccb9ab Testing bigger LLM config, referencing 'Attention Is All You Need' (pytorch) 2014-04-19 00:56:00 +02:00
saymrwulf
72ba352028 Trying out boneh-franklin approach for IBE (ref. 2003 paper) (pytorch) 2014-04-19 06:24:00 +02:00
saymrwulf
d83eaf251e Quick fix, referencing a known issue from the official repo (pytorch) 2014-04-19 04:05:00 +02:00
saymrwulf
525e69c6f1 Refactor for clarity, might break a few tests though (pytorch) 2014-04-03 23:10:00 +02:00
saymrwulf
a4e677ef16 Late-night bugfix on financial RL environment (pytorch) 2014-04-03 06:23:00 +02:00
saymrwulf
fca6d22957 Late-night bugfix on financial RL environment (pytorch) 2014-04-03 07:39:00 +02:00
saymrwulf
99267bd9dc Testing bigger LLM config, referencing 'Attention Is All You Need' (pytorch) 2014-04-03 01:39:00 +02:00
saymrwulf
bebf513ac6 Refactor for clarity, might break a few tests though (pytorch) 2014-02-01 23:18:00 +01:00
saymrwulf
d4942d7f6e Implementing approach from a new paper read last night (pytorch) 2014-02-01 02:35:00 +01:00
saymrwulf
06d57552a7 Refactor for clarity, might break a few tests though (pytorch) 2014-02-01 06:23:00 +01:00
saymrwulf
6750eb3298 Quick fix, referencing a known issue from the official repo (pytorch) 2014-01-30 22:47:00 +01:00
saymrwulf
aad0ff916b Quick fix, referencing a known issue from the official repo (pytorch) 2014-01-24 04:16:00 +01:00
saymrwulf
fababcc7e1 Experimenting with FPGA constraints (source: Trimberger 'Three Ages of FPGAs') (pytorch) 2014-01-24 23:53:00 +01:00
saymrwulf
60cffecc0a Quick fix, referencing a known issue from the official repo (pytorch) 2014-01-24 07:05:00 +01:00
saymrwulf
257fb73521 Implementing approach from a new paper read last night (pytorch) 2014-01-07 23:21:00 +01:00
saymrwulf
0e0e0a075e Implementing approach from a new paper read last night (pytorch) 2014-01-07 18:58:00 +01:00
saymrwulf
512be0049c Quick fix, referencing a known issue from the official repo (pytorch) 2014-01-01 05:53:00 +01:00
saymrwulf
5980b1984b Experimenting with FPGA constraints (source: Trimberger 'Three Ages of FPGAs') (pytorch) 2014-01-01 19:32:00 +01:00
saymrwulf
8a4a2cbf80 Trying out boneh-franklin approach for IBE (ref. 2003 paper) (pytorch) 2014-01-01 18:22:00 +01:00
saymrwulf
db3f5260e5 Quick fix, referencing a known issue from the official repo (pytorch) 2014-01-01 08:34:00 +01:00
Ting Lu
d0e70c4fd3 windows Magma build for cu128 (#146653) 
https://github.com/pytorch/pytorch/issues/145570

removing `.ci/pytorch/windows/internal/cuda_install.bat` as it is a duplicate with` .github/scripts/windows/cuda_install.bat`. The later one is the one in use - https://github.com/pytorch/pytorch/pull/146653/files#diff-613791f266f2f7b81148ca8f447b0cd6c6544f824f5f46a78a2794006c78957bR8

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146653
Approved by: https://github.com/atalman

Co-authored-by: atalman <atalman@fb.com>
 2025-02-10 13:48:55 +00:00
Tom Ritchford
6f15a609d3 Test typing of arithmetic operators on Tensor (see #145838) (#146426) 
See #145838

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146426
Approved by: https://github.com/Skylion007
 2025-02-10 12:19:56 +00:00
Jack Taylor
c24038025d [ROCm] Unskip std:bad_alloc failures (#146407) 
Flakey MI300 issue related to memory usage should now be resolved after https://github.com/pytorch/pytorch/actions/runs/13007160888?pr=145829.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146407
Approved by: https://github.com/jeffdaily
 2025-02-10 11:01:56 +00:00
yousoumar
c88ae00692 fix: replace stderr with stdout for download messages in hub.py (#146475) 
This PR addresses an issue where download logs in `hub.py` are sent to `stderr` instead of `stdout`. Hence, when running models with workers, these messages are incorrectly categorized as errors, leading to confusion.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/146475
Approved by: https://github.com/mikaylagawarecki
 2025-02-10 10:46:10 +00:00
gasoonjia
6667e5d786 [dim order] solve broken doc (#146641) 
Differential Revision: [D69265340](https://our.internmc.facebook.com/intern/diff/D69265340/)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/146641
Approved by: https://github.com/svekars, https://github.com/Jack-Khuu
 2025-02-10 07:51:26 +00:00
Xilun Wu
c4d835fbab [DTensor][conv] add DTensor convolution_backward op support for case where the input Tensor has requires_grad=False (#142278) 
Fixes #142058

## Summary
DTensor `convolution_backward` op throws exception when the input Tensor has `requires_grad=False` which happens if the conv layer is the first layer in the model.

ATEN convolution_backward op Usually returns 3 Tensors (grad_input, grad_weight, grad_bias) and the `grad_input` is actually an Optional[Tensor] which can be `None` in the case mentioned above.

However, the DTensor sharding propagation rule and corresponding TP conv backward implementation both assume that the `grad_input` would be existent.

## Fix
allow the `grad_input` to be `None` for `convolution_backward` op.

## Test
`pytest test/distributed/tensor/test_convolution_ops.py`

## Follow-up
The current implementation of DTensor conv op also ignores `output_mask` and this may need further care.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/142278
Approved by: https://github.com/bdhirsh
 2025-02-10 07:06:40 +00:00
Ke Wen
effc545274 [DDP] Use NCCL allocated memory for gradient bucket (#146589) 
So that NVLink SHARP comes with zero-copy on H100+ platforms, for DDP applications.
Less SM usage, less memory contention between NCCL kernel and compute kernels.

Added env `DDP_DISABLE_COMM_MEM` as a back-out option:
```
An environment variable to disable comm-optimized memory pool.
Default is 0, which means comm-optimized memory pool is enabled.
Users can set it to 1 in case of seeing regression or OOM (because this
comm MemPool may not share space with regular compute MemPool).
```

Differential Revision: [D69297766](https://our.internmc.facebook.com/intern/diff/D69297766)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/146589
Approved by: https://github.com/syed-ahmed, https://github.com/c-p-i-o, https://github.com/fduwjj
 2025-02-10 05:23:11 +00:00
Simon Fan
387c993c3b [ca] remove private API: _compiled_autograd_should_lift (#146720) 
Since the functional autograd + compiled autograd migration, we don't trace into nodes anymore, and everything is lifted. We can't support this flag which tries to inline make_fx style in CA initial pass. There's no more usage internally.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/146720
Approved by: https://github.com/zou3519
 2025-02-10 04:29:57 +00:00
zeshengzong
e8304f08fe Fix torch.take_along_dim param type and default description (#146474) 
## Changes

- Change type description to `LongTensor`, consistent with [`torch.take`](https://pytorch.org/docs/stable/generated/torch.take.html)
- Add `dim` param default value description

## Test Result

**Before**
![image](https://github.com/user-attachments/assets/720ce158-2bc1-48b5-a188-56fcc7188d96)

**After**
![image](https://github.com/user-attachments/assets/05fe20bd-9476-4b97-ac2b-9b161d6532a1)

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146474
Approved by: https://github.com/mikaylagawarecki
 2025-02-10 01:19:30 +00:00
Simon Fan
298226f358 [dynamo] check for incompatible configs (#146513) 
internal: https://fb.workplace.com/groups/1075192433118967/permalink/1599802033991335/

Assuming flags don't change during compilation, we shouldn't allow incompatible configs to be set at torch.compile wrap time.

Not in this PR: For flags that need to change during compilation, we'd have to be strict about where they can be used in the compile lifecycle

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146513
Approved by: https://github.com/williamwen42

Co-authored-by: Gabriel Ferns <gabeferns@meta.com>
 2025-02-10 00:44:23 +00:00
Davide Italiano
2a55311773 [cuda] Simplify the sinc function a bit. (#146774) 
`else` after `return` can be removed & the indentation can be reduced, for readability.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/146774
Approved by: https://github.com/malfet
 2025-02-09 20:09:34 +00:00
drisspg
b133907d0a Update strided test to float32 (#146748) 
Fixes #146377

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146748
Approved by: https://github.com/BoyuanFeng, https://github.com/leijurv
 2025-02-09 17:41:35 +00:00
Davide Italiano
91c4bf39d3 [mps] Add a shader for spherical_bessel_j0. (#146771) 
In preparation for adding the operation to inductor/eager.
Adapted from the CUDA version of the shader.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/146771
Approved by: https://github.com/malfet
 2025-02-09 05:11:17 +00:00