KimmiShi
|
98c96e790b
Update flops profiler to handle attn and __matmul__ (#4724)
|
8 月之前 |
Logan Adams
|
61391229c9
Update flops profiler to recurse (#4374)
|
11 月之前 |
Logan Adams
|
a25a67a083
Fix scale factor on flops profiler (#4500)
|
1 年之前 |
Alexander Jipa
|
b354c28b76
polishing timers and log_dist (#3996)
|
1 年之前 |
Pinstripe Potoroo
|
f4d18fa2b7
fix gates size retrieval logic in _rnn_flops (#3921)
|
1 年之前 |
Alexander Jipa
|
488a1b9879
fixing flops profiler formatting, units and precision (#3927)
|
1 年之前 |
Pinstripe Potoroo
|
3491e32d72
fix rnn flop profiler to compute flops instead of macs (#3833)
|
1 年之前 |
Pinstripe Potoroo
|
c1c1d2496f
fix retrieval of out_channels in _conv_trans_flops_compute (#3834)
|
1 年之前 |
Guorun
|
24c7d7f14a
use `Flops Profiler` to test `model.generate()` (#2515)
|
1 年之前 |
Cheng Li
|
a76cced3fa
fix interpolate flops compute (#3782)
|
1 年之前 |
Bill Luo
|
062408683c
[Fix] _conv_flops_compute when padding is a str and stride=1 (#3169)
|
1 年之前 |
Alejandro Dubrovsky
|
9a8f6a1d3c
Account for expert parameters when calculating the total number of parameters in the model (#3720)
|
1 年之前 |
Guo Yejun
|
460bec4679
flops_profiler: add option recompute_fwd_factor for the case of activation recompute (#3362)
|
1 年之前 |
Michael Wyatt
|
b361c72761
Update DeepSpeed copyright license to Apache 2.0 (#3111)
|
1 年之前 |
Jeff Rasley
|
91d63e0228
update formatter version and style settings (#3098)
|
1 年之前 |
Jeff Rasley
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
Cheng Li
|
b5750b6449
fix upsample flops compute by skipping unused kargs (#2773)
|
1 年之前 |
swli
|
e2a31d800b
some fix in flops_profiler (#2068)
|
1 年之前 |
Ma, Guokai
|
98cc35b6a8
Abstract accelerator (step 3) (#2677)
|
1 年之前 |
Jeff Rasley
|
2110552167
Fixes for torch 1.14 due to new torch.numel return type (#2522)
|
1 年之前 |
Cheng Li
|
8da0238b7a
rollback ds config changes (#2395)
|
2 年之前 |
Cheng Li
|
5d1f595c94
update pytorch pool operator function signiture (#2443)
|
2 年之前 |
Zion Wu
|
6bfcf3c694
Fix wrong unit of latency in flops-profiler (#2090) (#2095)
|
2 年之前 |
Alex Hedges
|
316c4a43e0
Add flake8 to pre-commit checks (#2051)
|
2 年之前 |
Aman Sanger
|
9027f861f2
Dont overwrite hook handles in flop profiler (#2106)
|
2 年之前 |
Cheng Li
|
0ad08608a5
remove require grad in params count (#2065)
|
2 年之前 |
Michael Wyatt
|
5218177922
fixed print statement (#2038)
|
2 年之前 |
Cheng Li
|
6719b46bd8
fix typo when getting kernel dim in conv calculation (#1989)
|
2 年之前 |
TongXU
|
d8ed3ce445
remove the print line in _einsum_flops_compute function (#1885)
|
2 年之前 |
Cheng Li
|
2151c787a2
Generalize the model input format of the flops profiler get_model_profile() API (#1768)
|
2 年之前 |