strisunshinewentingwang/pytorch

Pull Requests Commits

[functorch] improve coverage testing

zou3519 committed 2 years ago

4e3b1503

[functorch] hessian API

zou3519 committed 2 years ago

cdb56a0e

[functorch] exported register_decomposition to be public

Chillee committed 2 years ago

4e75634a

[functorch] Fix edge case with in-place captures and grad transforms (pytorch/functorch#272)

zou3519 committed 2 years ago

f1c9426c

[functorch] Revert "Making some things kwarg only"

zou3519 committed 2 years ago

21a4c9fb

[functorch] Making some things kwarg only

zou3519 committed 2 years ago

37123cd2

[functorch] Strict mode for jvp

zou3519 committed 2 years ago

30f787c6

[functorch] Fix unrelated output behavior for jvp

zou3519 committed 2 years ago

b8e25d25

[functorch] jacfwd supports multiple inputs and multiple outputs (pytorch/functorch#270)

zou3519 committed 2 years ago

1b10fb4f

[functorch] jvp transform accepts functions that take multiple inputs and outputs (pytorch/functorch#267)

zou3519 committed 2 years ago

e19546fb

[functorch] jacrev, multiple outputs support (pytorch/functorch#265)

zou3519 committed 2 years ago

403e2962

[functorch] Added im2col batch rule and enabled vmap for nn.functional.unfold op (pytorch/functorch#262)

vfdev-5 committed 2 years ago

5a58d5f7

[functorch] changed naming from aot autograd to eager compilation

Chillee committed 2 years ago

8769c9a2

[functorch] updated some eager compilation code

Chillee committed 2 years ago

1eaf24c9

[functorch] [CompileCache] Adding compilation cache (pytorch/functorch#250)

anijain2305 committed 2 years ago

91295695

[functorch] Document some silently uncorrect behavior before I forget

zou3519 committed 2 years ago

e6a1dfbd

[functorch] Skip testing empty_like instead of xfail (pytorch/functorch#259)

zou3519 committed 2 years ago

752d0ed8

[functorch] quick fix ci

zou3519 committed 2 years ago

7142acbf

[functorch] Update functorch lagging op db (pytorch/functorch#261)

zou3519 committed 2 years ago

41699bb9

[functorch] Better error on item() and data dependent control flow

zou3519 committed 2 years ago

9962b908

[functorch] [AOT Autograd] refactor partitioner to extract arbitrary graph given inputs and outputs (pytorch/functorch#258)

Chillee committed 2 years ago

1321ed04

[functorch] Added batch rule for (adaptive_)avg_poolNd (pytorch/functorch#248)

vfdev-5 committed 2 years ago

bef46036

[functorch] Link to docs site

zou3519 committed 2 years ago

b485247d

[functorch] fix softplus_backward decomposition

Chillee committed 2 years ago

d70f21b6

[functorch] Fixed nn.functional.pad constant mode (pytorch/functorch#249)

vfdev-5 committed 2 years ago

eb9ba238

[functorch] Added backward batch rule for pad replicate/reflect modes (pytorch/functorch#251)

vfdev-5 committed 2 years ago

82f8f532

[functorch] Enables tests for index_put_ on CUDA (pytorch/functorch#252)

vfdev-5 committed 2 years ago

f3f8d8bd

[functorch] update op analysis

zou3519 committed 2 years ago

126b4b91

[functorch] Refactored BatchedTensorImpl to have single bdim and level (pytorch/functorch#239)

vfdev-5 committed 2 years ago

d9c4c43e

[functorch] Add instructions on how to do the functorch docs deploy

zou3519 committed 2 years ago

5cd8af24

Newer Older