JointGraph-based Training Prototype #1794
Open
+427
−3
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
meta-cla
bot
added
the
CLA Signed
tianyu-l
requested changes
Oct 9, 2025
SherlockNoMad
commented
Oct 10, 2025
Comment on lines
+296
to
+307
| # Hack: convert args and kwargs to DTensor. This should be fixed at data loader. | ||
| # This works, but kinda cheating? | ||
| dt_args = tuple( | ||
| DTensor.from_local(arg, self.parallel_dims.world_mesh["tp"], [Replicate()]) | ||
| for arg in args | ||
| ) | ||
|
|
||
| # RuntimeError('Sharding propagation failed for Op(op=aten.embedding.default, args_schema=Spec(S(0) on (2048, 256)), Spec((Shard(dim=0), Replicate()) on (16, 2048)) @ mesh: (2, 4))') | ||
| # dt_args = tuple(DTensor.from_local(arg, self.parallel_dims.world_mesh, [Shard(0), Replicate()]) for arg in args) | ||
|
|
||
| # RuntimeError('Sharding propagation failed for Op(op=aten.embedding.default, args_schema=Spec(S(0) on (2048, 256)), Spec(S(0) on (16, 2048)) @ mesh: (2,))') | ||
| # dt_args = tuple(DTensor.from_local(arg, self.parallel_dims.world_mesh["dp_shard"], [Shard(0)]) for arg in args) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is an e2e prototype to run llama3-simplefsdp using export-y aot_autograd workflow.
Setup: shard_dp = 2, tp = 4.
MVP
aot_compile_joint_with_descriptorsIssues
Repro steps:
NGPU=8 CONFIG_FILE="./torchtitan/models/llama3/train_configs/debug_model.toml" with-proxy ./run_train.sh --model.name joint_graph_runner.llama3 --compile.enable --parallelism.data_parallel_shard_degree=2 --parallelism.tensor_parallel_degree=4
Run with FlexAttention:
NGPU=8 CONFIG_FILE="./torchtitan/models/llama3/train_configs/debug_model.toml" with-proxy ./run_train.sh --model.name joint_graph_runner.llama3 --compile.enable --parallelism.data_paral
lel_shard_degree=2 --parallelism.tensor_parallel_degree=4 --model.flavor=debugmodel_flex_attn
Sample output:
P1975157784: rank0_autograd_function_0fea2786.py
P1975158481: rank1_autograd_function_28587623.py