[PyTorch][torch.compile] Replace TensorProto with make_empty_traceable - #13

Open

kshitij12345 wants to merge 28 commits into

pggPL:linear_compilefrom

kshitij12345:linear_compile_no_proto

Commits on Jun 29, 2026

[PyTorch] Make tensorless quantizers opaque value objects for torch.compile
pggPL
committed
f3401df
[PyTorch] Drop quantizer value registry; reconstruct via __fx_repr__ globals
pggPL
committed
c4ad54c
[PyTorch] Split dynamo.py into a dynamo/ package
pggPL
committed
a06324b
[PyTorch] Raise in quantizer __fx_repr__ when a process group is stored
pggPL
committed
ea5b396
[PyTorch] Cover NVFP4 in quantizer value-object test
pggPL
committed
aa65e34
Reject a value quantizer that carries an amax reduction group in __eq__/__hash__

pggPL
and
claude
committed
e1b1db6
Recognize value-opaque quantizers via a class flag

pggPL
and
claude
committed
8c33d0e
Address review: narrow opaque-type except, add fullgraph test, fix nvfp4 value key

pggPL
and
claude
committed
945f62d
Restore NVFP4 rht_matrix on value-key rebuild; assert quantize round-trip

pggPL
and
claude
committed
e3c8f43
Enforce process-group rejection in _value_key, not __fx_repr__; add test

pggPL
and
claude
committed
3f68621
Strengthen fullgraph test: quantize/dequantize via a custom op, not passthrough

pggPL
and
claude
committed
32d1768
Clarify comments: rht_matrix_random_sign_mask_t derivation; why the opaque flag

pggPL
and
claude
committed
28bde9e
Reword opaque-flag comment: self-contained, no Linear reference

pggPL
and
claude
committed
2c3c5df
Cover is_opaque_value_type with the import-safety guard too

pggPL
and
claude
committed
826f271
Add TensorProto mechanism for data-free quantized tensor allocation

pggPL
and
claude
committed
ad1ccce
[PyTorch] torch.compile: dedup cached FP8 weight from saved-for-backward

pggPL
and
claude
committed
ea3df7a
[PyTorch] nvfp4: emit _describe_buffers in canonical flatten order

pggPL
and
claude
committed
4997929
Address review: error on undescribed buffers, gate nvfp4 test on HW support

pggPL
and
claude
committed
50c11cd
[PyTorch] Workaround torch.compile staticmethod guard bug in NVFP4 _describe_buffers

pggPL
and
claude
committed
ff48e52

Commits on Jun 30, 2026

[PyTorch] torch.compile: wrap pybind11 UB methods as compile-time constants; fix SP memory leak; test suite hook-up

pggPL
and
claude
committed
e1e271c
[pre-commit.ci] auto fixes from pre-commit.com hooks

pre-commit-ci[bot]
authored and
pggPL
committed
598a07c
Provide explicit QuantizerRoles in torch.compile custom-recipe test
pggPL
committed
05af2a0
Add torch.compile custom-op path for Linear

pggPL
and
claude
committed
9bd16fd
[PyTorch] torch.compile: register TE custom ops via torch.library.custom_op

pggPL
and
claude
committed
fdac659
[PyTorch] custom_op: note pytorch/pytorch#187434 enables dropping None sentinel

pggPL
and
claude
committed
554f5a8

Commits on Jul 2, 2026

Commits on Jul 3, 2026

[PyTorch][test] Fix NVFP4 value-object test: set with_post_rht_amax=True

kshitij12345
and
claude
committed
a582eb5

Navigation Menu

[PyTorch][torch.compile] Replace TensorProto with make_empty_traceable - #13

[PyTorch][torch.compile] Replace TensorProto with make_empty_traceable#13
kshitij12345 wants to merge 28 commits into
pggPL:linear_compilefrom
kshitij12345:linear_compile_no_proto

Commits on Jun 29, 2026

[PyTorch] Make tensorless quantizers opaque value objects for torch.compile

[PyTorch] Drop quantizer value registry; reconstruct via __fx_repr__ globals

[PyTorch] Split dynamo.py into a dynamo/ package

[PyTorch] Raise in quantizer __fx_repr__ when a process group is stored

[PyTorch] Cover NVFP4 in quantizer value-object test

Reject a value quantizer that carries an amax reduction group in eq/hash

Recognize value-opaque quantizers via a class flag

Address review: narrow opaque-type except, add fullgraph test, fix nvfp4 value key

Restore NVFP4 rht_matrix on value-key rebuild; assert quantize round-trip

Enforce process-group rejection in _value_key, not __fx_repr__; add test

Strengthen fullgraph test: quantize/dequantize via a custom op, not passthrough

Clarify comments: rht_matrix_random_sign_mask_t derivation; why the opaque flag

Reword opaque-flag comment: self-contained, no Linear reference

Cover is_opaque_value_type with the import-safety guard too

Add TensorProto mechanism for data-free quantized tensor allocation

[PyTorch] torch.compile: dedup cached FP8 weight from saved-for-backward

[PyTorch] nvfp4: emit _describe_buffers in canonical flatten order

Address review: error on undescribed buffers, gate nvfp4 test on HW support

[PyTorch] Workaround torch.compile staticmethod guard bug in NVFP4 _describe_buffers

Commits on Jun 30, 2026

[PyTorch] torch.compile: wrap pybind11 UB methods as compile-time constants; fix SP memory leak; test suite hook-up

[pre-commit.ci] auto fixes from pre-commit.com hooks

Provide explicit QuantizerRoles in torch.compile custom-recipe test

Add torch.compile custom-op path for Linear

[PyTorch] torch.compile: register TE custom ops via torch.library.custom_op

[PyTorch] custom_op: note pytorch/pytorch#187434 enables dropping None sentinel

Commits on Jul 2, 2026

[PyTorch][torch.compile] Replace TensorProto with make_empty_traceable

[PyTorch][torch.compile] Guard NVFP4Quantizer.copy() tensor access under is_compiling

Commits on Jul 3, 2026

[PyTorch][test] Fix NVFP4 value-object test: set with_post_rht_amax=True