[BE] Lr schduler flatten (#794)

Currently, lr_scheduler is stored differently as optimizer, model and
data_loader, with keys to be "lr_scheduler_0", "lr_scheduler_1", ...
stored in the state
This PR aims to flatten lr_shceduler so that all the schedulers would be
stored as a list of state_dict under self.state['lr_scheduler'], which
is consistent with optimizer

Here we have the assumption that all the optimziers have the same
lr_scheduler, thus only to save a single lr_scheduler's state_dict and
load it to all the schedulers works here.

The lr_scheduler has the state_dict like:
`{'base_lrs': [0.0003], 'last_epoch': 1, 'verbose': False,
'_step_count': 2, '_get_lr_called_within_step': False, '_last_lr':
[2.985074626865671e-06], 'lr_lambdas': [{}]}`

The PR is tested by 2 parts:
1. test lr_scheduler value before and after checkpoint, resharding with
degree changes on tp and pp.
[dp=2, tp=4, pp=1] -> [dp=2, tp=1, pp=4]
[dp=2, tp=1, pp=4] -> [dp=2, tp=4, pp=1]
date_loader does not support resharding right now.

logs:
[dp=2, tp=4, pp=1] 
step 5 before saving to checkpoint:
[{'lr': 8.955223880597014e-06, ...}]

step 10 after loading from checkpoint and reshard to [dp=2, tp=2, pp=2]:
[{'lr': 1.6417910447761194e-05, ...}, {'lr': 1.6417910447761194e-05,
...}]

[dp=8, tp=1, pp=1]
step 5 without checkpoint:
[{'lr': 8.955223880597014e-06, ...}]
step 10 without checkpoint:
[{'lr': 1.6417910447761194e-05, ...}]

2. Memory trace:
Before the flatten, rerun llama3_8b.toml from step 5 to step 10:
<img width="1166" alt="Screenshot 2025-01-16 at 2 40 03 PM"
src="https://github.com/user-attachments/assets/d3e84d63-30be-4604-823b-68bd217498a0"
/>

After the flatten, rerun llama3_8b.toml from step 5 to step 10:
<img width="1166" alt="Screenshot 2025-01-16 at 2 40 21 PM"
src="https://github.com/user-attachments/assets/b6ed68ae-2dbf-400a-b723-06eae6740ade"
/>

Author: mori360

torchtitan/checkpoint.py

@@ -183,9 +183,9 @@ def __init__(
                 "model": ModelWrapper(model_parts),
                 "optimizer": optimizers,
                 "dataloader": dataloader,
+                "lr_scheduler": lr_schedulers,
             }
         )
-        self.states.update(lr_schedulers.get_lr_scheduler_state())
 
         self.folder = os.path.join(job_config.job.dump_folder, ckpt_config.folder)
         self.interval_type = (

torchtitan/optimizer.py

@@ -178,7 +178,7 @@ def linear_warmup_linear_decay(
     return curr_adjustment
 
 
-class SchedulersContainer:
+class SchedulersContainer(Stateful):
     """Util for calling step on multiple learning rate schedulers needed for virtual pipeline stages"""
 
     def __init__(self, optimizers, lr_lambda) -> None:
@@ -190,16 +190,21 @@ def step(self) -> None:
         for scheduler in self.schedulers:
             scheduler.step()
 
-    def get_lr_scheduler_state(self) -> Dict[str, Any]:
-        state_dict = {}
-        if len(self.schedulers) == 1:
-            state_dict["lr_scheduler"] = self.schedulers[0]
-        else:
-            # For now, pipeline-parallel with looped schedules does not support resharding for lr_scheduler.
-            # It should only support saving and loading a distributed checkpoint with the same number of pp ranks
-            for idx, lr_scheduler in enumerate(self.schedulers):
-                state_dict[f"lr_scheduler_{idx}"] = lr_scheduler
-        return state_dict
+    def state_dict(self) -> Dict[str, Any]:
+        # Currently, we have one scheduler per optimizer. However, when using MultiSchedule PP or optimizer-in-backward,
+        # there are multiple optimizers and schedulers, but the scheduler state_dict remains the same for all.
+        # Therefore, we only save the first one and later load it for all.
+        assert (
+            len(self.schedulers) > 0
+        ), "Must have at least one scheduler to save state_dict"
+        return self.schedulers[0].state_dict()
+
+    def load_state_dict(self, state_dict: Dict[str, Any]) -> None:
+        # Load the same state_dict for all schedulers. The key value we're concerned with in scheduler.state_dict() is `last_epoch`,
+        # which is an integer that will be automatically copied. As long as `training.steps` and `training.warmup_steps` remain
+        # unchanged when resuming from a checkpoint, this approach is safe. We call `.copy()` here to ensure extra safety.
+        for scheduler in self.schedulers:
+            scheduler.load_state_dict(state_dict.copy())
 
 
 def build_lr_schedulers(optimizers, job_config: JobConfig) -> SchedulersContainer: