8000 Commits · yumin829928/torchrec · GitHub

More Web Proxy on the site http://driver.im/

Commits

Commits on May 17, 2024

EmbeddingsAllToOne world_size 1 case (#2014)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on May 15, 2024

Fix weights Optional[Tensor] issue (#2005)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on May 14, 2024

Fix CI with torchrec.pt2 (#1998)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on May 13, 2024

Fix Build Linux Wheels torchrec/pt2 (#1997)

PaulZhang12
authored and
facebook-github-bot
committed
Move multiprocess serialization tests into /fb/ir (#1975)

PaulZhang12
authored and
facebook-github-bot
committed
Permute export fix (#1985)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on May 10, 2024

Commits on May 9, 2024

Simplify Logic with KJT weights in QuantBatchedEmbeddingBag for TorchScript (#1956)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on May 1, 2024

Always return .int() conversion from unwrap_kjt

PaulZhang12
authored and
facebook-github-bot
committed
Create BoundsCheckMode fused_param for inference

PaulZhang12
authored and
facebook-github-bot
committed
Move eager model processing functions to torchrec/inference

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 26, 2024

Fix FeatureProcessor device only to meta

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 19, 2024

OSS serialization with dataclasses (#1860)

PaulZhang12
authored and
facebook-github-bot
committed
Fix deserialization of nested fqn + serialization returns fqns for preserve_module_signature (#1906)

PaulZhang12
authored and
facebook-github-bot
committed
Fix pytree flatten/unflatten for KeyedTensor (#1899)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 17, 2024

Tests for torch.export sharded inference module (#1885)

PaulZhang12
authored and
facebook-github-bot
committed
Fix FP copying in Sharded Quant FPEBC (#1881)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 16, 2024

Fixing device issues (#1880)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 10, 2024

KJT permute torch export support (#1850)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 9, 2024

Use "nn_module_stack" of ExportedProgram for type information in serialization/deserialization (#1856)

PaulZhang12
authored and
facebook-github-bot
committed
Fix OSS precommit CI (#1857)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 8, 2024

VBE training benchmarks (Manual) (#1855)

PaulZhang12
authored and
facebook-github-bot
committed
Optimize VBE input generation (#1854)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Apr 5, 2024

Introduce Serializer in TorchRec for torch.export PEA (#1848)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Mar 22, 2024

KJT split torch export support (#1816)

PaulZhang12
authored and
facebook-github-bot
committed
KJT __getitem__ torch export support (#1815)

PaulZhang12
authored and
facebook-github-bot
committed
More efficient cumsum for offset_per_key calculation for torch.export (#1788)

PaulZhang12
authored and
facebook-github-bot
committed
Re-enable on diff as benchmarks have improved and capactiy increases (#1821)

PaulZhang12
authored and
facebook-github-bot
committed
Benchmark subprocess sleep for multiprocessing queue fix (#1820)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Mar 21, 2024

Disable CW in manual script due to profiling issue (#1818)

PaulZhang12
authored and
facebook-github-bot
committed
Fix benchmark bug with optimization (#1817)

PaulZhang12
authored and
facebook-github-bot
committed

Commits on Mar 20, 2024

5x speedup of benchmarks with input generation (#1813)

PaulZhang12
authored and
facebook-github-bot
committed
Disable Traces on Benchmark when in ServiceLab (#1804)

PaulZhang12
authored and
facebook-github-bot
committed

0