Sherlock Huang

API for checking is_canonical

2

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #83134

Preserve node's stack trace during retrace

2

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #83050 AOTAutograd retraces graph module produced by torch dynamo, this PR preserves the stack trace in the original fx.Node.

Improve and expose cpp_backtrace to python binding

11

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #84896 We can now get cpp stack trace by calling torch.utils.get_cpp_backtrace() Sample output when calling from a torch_dispatch stack: ``` frame #23:...

[NOT4LAND] Initial draft of Export's logical schema

7

Draft for PT2 Export Schema. This is the logical representation of schema. Actual schema will be rewritten with FlatBuffer or other serialization lib. An example of readable format of ResNet18...

Consistent meta reg across dispatcher, py-dispatcher, active_meta_table

4

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #87662

Add sym_size/stride/numel/storage_offset to native_function.yaml

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #91919

[Compiler Toolkit] Enable nested_compile_region on TransformerBlock

1

Need to run with fix in https://github.com/pytorch/pytorch/pull/166702 ``` NGPU=8 CONFIG_FILE=./torchtitan/models/llama3/train_configs/debug_model.toml ./run_train.sh --model.name compiler_toolkit.llama3 --parallelism.data_parallel_shard_degree=2 --parallelism.tensor_parallel_degree=4 ``` Current output: P2016557983 Observations - I see each TransformerBlock becomes one subgraph, look for...

[Compiler Toolkit] Assert compile.enable=False

3

As title, compile.enable should be False in the compiler toolkit style workflow