cleanup update method

src/lerobot/scripts/lerobot_train.py

@@ -16,14 +16,13 @@
 import logging
 import os
 import time
-from collections.abc import Callable
 from contextlib import nullcontext
 from pprint import pformat
 from typing import Any
 
 import torch
+from accelerate import Accelerator
 from termcolor import colored
-from torch.amp import GradScaler
 from torch.optim import Optimizer
 
 from lerobot.configs import parser
@@ -36,7 +35,6 @@ from lerobot.envs.utils import close_envs
 from lerobot.optim.factory import make_optimizer_and_scheduler
 from lerobot.policies.factory import make_policy, make_pre_post_processors
 from lerobot.policies.pretrained import PreTrainedPolicy
-from lerobot.policies.utils import get_device_from_parameters
 from lerobot.rl.wandb_utils import WandBLogger
 from lerobot.scripts.lerobot_eval import eval_policy_all
 from lerobot.utils.logging_utils import AverageMeter, MetricsTracker
@@ -50,10 +48,8 @@ from lerobot.utils.train_utils import (
 )
 from lerobot.utils.utils import (
     format_big_number,
-    get_safe_torch_device,
     has_method,
     init_logging,
-    is_launched_with_accelerate,
 )
 
 
@@ -63,17 +59,15 @@ def update_policy(
     batch: Any,
     optimizer: Optimizer,
     grad_clip_norm: float,
-    grad_scaler: GradScaler,
+    accelerator: Accelerator,
     lr_scheduler=None,
-    use_amp: bool = False,
     lock=None,
-    accelerator: Callable | None = None,
 ) -> tuple[MetricsTracker, dict]:
     """
     Performs a single training step to update the policy's weights.
 
     This function executes the forward and backward passes, clips gradients, and steps the optimizer and
-    learning rate scheduler. It also handles mixed-precision training via a GradScaler.
+    learning rate scheduler. Accelerator handles mixed-precision training automatically.
 
     Args:
         train_metrics: A MetricsTracker instance to record training statistics.
@@ -81,11 +75,9 @@ def update_policy(
         batch: A batch of training data.
         optimizer: The optimizer used to update the policy's parameters.
         grad_clip_norm: The maximum norm for gradient clipping.
-        grad_scaler: The GradScaler for automatic mixed-precision training.
+        accelerator: The Accelerator instance for distributed training and mixed precision.
         lr_scheduler: An optional learning rate scheduler.
-        use_amp: A boolean indicating whether to use automatic mixed precision.
         lock: An optional lock for thread-safe optimizer updates.
-        accelerator: An optional accelerator, for multi-gpu training.
 
     Returns:
         A tuple containing:
@@ -93,37 +85,25 @@ def update_policy(
         - A dictionary of outputs from the policy's forward pass, for logging purposes.
     """
     start_time = time.perf_counter()
-    device = get_device_from_parameters(policy)
     policy.train()
+    
     # Let accelerator handle mixed precision
-    with accelerator.autocast() if accelerator else (torch.autocast(device_type=device.type) if use_amp else nullcontext()):
+    with accelerator.autocast():
         loss, output_dict = policy.forward(batch)
         # TODO(rcadene): policy.unnormalize_outputs(out_dict)
 
-    if accelerator:
+    # Use accelerator's backward method
-        accelerator.backward(loss)
+    accelerator.backward(loss)
-        if grad_clip_norm > 0:
+    
-            grad_norm = accelerator.clip_grad_norm_(policy.parameters(), grad_clip_norm)
+    # Clip gradients if specified
-        else:
+    if grad_clip_norm > 0:
-            grad_norm = torch.tensor(0.0, device=policy.device)
+        grad_norm = accelerator.clip_grad_norm_(policy.parameters(), grad_clip_norm)
-        optimizer.step()
     else:
-        grad_scaler.scale(loss).backward()
+        grad_norm = torch.tensor(0.0, device=accelerator.device)
-        # Unscale the gradient of the optimizer's assigned params in-place **prior to gradient clipping**.
-        grad_scaler.unscale_(optimizer)
     
-        grad_norm = torch.nn.utils.clip_grad_norm_(
+    # Optimizer step
-            policy.parameters(),
+    with lock if lock is not None else nullcontext():
-            grad_clip_norm,
+        optimizer.step()
-            error_if_nonfinite=False,
-        )
-
-        # Optimizer's gradients are already unscaled, so scaler.step does not unscale them,
-        # although it still skips optimizer.step() if the gradients contain infs or NaNs.
-        with lock if lock is not None else nullcontext():
-            grad_scaler.step(optimizer)
-        # Updates the scale for next iteration.
-        grad_scaler.update()
     
     optimizer.zero_grad()
 
@@ -131,13 +111,9 @@ def update_policy(
     if lr_scheduler is not None:
         lr_scheduler.step()
 
-    if accelerator:
+    # Update internal buffers if policy has update method
-        if has_method(accelerator.unwrap_model(policy, keep_fp32_wrapper=True), "update"):
+    if has_method(accelerator.unwrap_model(policy, keep_fp32_wrapper=True), "update"):
-            accelerator.unwrap_model(policy, keep_fp32_wrapper=True).update()
+        accelerator.unwrap_model(policy, keep_fp32_wrapper=True).update()
-    else:
-        if has_method(policy, "update"):
-            # To possibly update an internal buffer (for instance an Exponential Moving Average like in TDMPC).
-            policy.update()
 
     train_metrics.loss = loss.item()
     train_metrics.grad_norm = grad_norm.item()

src/lerobot/utils/utils.py

@@ -177,21 +177,6 @@ def format_big_number(num, precision=0):
 
     return num
 
-
-def is_launched_with_accelerate() -> bool:
-    """Check if the script was launched in a distributed training context.
-    
-    This checks for standard distributed training environment variables that are set
-    by accelerate launch, torchrun, or torch.distributed.launch.
-    
-    Returns:
-        True if running in a distributed context, False otherwise.
-    """
-    # Check for LOCAL_RANK which is the standard way to detect distributed training
-    # This is set by accelerate, torchrun, and torch.distributed.launch
-    return "LOCAL_RANK" in os.environ
-
-
 def say(text: str, blocking: bool = False):
     system = platform.system()