fix: overfit_batches uses same batch for train and val

docs/source-pytorch/common/trainer.rst

@@ -759,6 +759,9 @@ overfit_batches
 Uses this much data of the training & validation set.
 If the training & validation dataloaders have ``shuffle=True``, Lightning will automatically disable it.
 
+* When set to exactly 1, the same batch is used for both training and validation steps, which is useful for debugging model implementation
+* For other values, sequential sampling (no shuffling) is used
+
 Useful for quickly debugging or trying to overfit on purpose.
 
 .. testcode::
@@ -769,9 +772,13 @@ Useful for quickly debugging or trying to overfit on purpose.
     # use only 1% of the train & val set
     trainer = Trainer(overfit_batches=0.01)
 
-    # overfit on 10 of the same batches
+    # overfit on 10 (same) train batches & 10 (same) val batches
     trainer = Trainer(overfit_batches=10)
 
+    # debug by training and validating on exactly the same single batch
+    # (useful for verifying model implementation)
+    trainer = Trainer(overfit_batches=1)
+
 plugins
 ^^^^^^^
 

src/lightning/pytorch/trainer/connectors/data_connector.py

@@ -244,19 +244,64 @@ def _get_distributed_sampler(
 
 
 def _resolve_overfit_batches(combined_loader: CombinedLoader, mode: RunningStage) -> None:
+    """Resolve overfit batches by ensuring the same batch is used for both training and validation."""
     all_have_sequential_sampler = all(
         isinstance(dl.sampler, SequentialSampler) for dl in combined_loader.flattened if hasattr(dl, "sampler")
     )
     if all_have_sequential_sampler:
         return
+
     rank_zero_warn(
         f"You requested to overfit but enabled {mode.dataloader_prefix} dataloader shuffling."
         f" We are turning off the {mode.dataloader_prefix} dataloader shuffling for you."
     )
-    updated = [
-        _update_dataloader(dl, sampler=SequentialSampler(dl.dataset), mode=mode) if hasattr(dl, "dataset") else dl
-        for dl in combined_loader.flattened
-    ]
+
+    # Get the first batch from the training dataloader
+    first_batch = None
+    if mode == RunningStage.TRAINING:
+        for dl in combined_loader.flattened:
+            if hasattr(dl, "dataset"):
+                first_batch = next(iter(dl))
+                break
+
+    # Create new dataloaders with SequentialSampler
+    updated = []
+    for dl in combined_loader.flattened:
+        if hasattr(dl, "dataset"):
+            if mode == RunningStage.VALIDATING and first_batch is not None:
+                # For validation, create a custom sampler that always returns the first batch
+                class SingleBatchSampler(Sampler):
+                    def __init__(self, batch):
+                        self.batch = batch
+
+                    def __iter__(self):
+                        yield self.batch
+
+                    def __len__(self):
+                        return 1
+
+                sampler = SingleBatchSampler(first_batch)
+            else:
+                sampler = SequentialSampler(dl.dataset)
+
+            # Create a new dataloader with the new sampler
+            dl = DataLoader(
+                dataset=dl.dataset,
+                batch_size=dl.batch_size,
+                sampler=sampler,
+                num_workers=dl.num_workers,
+                collate_fn=dl.collate_fn,
+                pin_memory=dl.pin_memory,
+                drop_last=dl.drop_last,
+                timeout=dl.timeout,
+                worker_init_fn=dl.worker_init_fn,
+                multiprocessing_context=dl.multiprocessing_context,
+                generator=dl.generator,
+                prefetch_factor=dl.prefetch_factor,
+                persistent_workers=dl.persistent_workers,
+            )
+        updated.append(dl)
+
     combined_loader.flattened = updated
 
 

tests/tests_pytorch/conftest.py

@@ -95,6 +95,12 @@ def restore_env_variables():
         "TF_GRPC_DEFAULT_OPTIONS",
         "XLA_FLAGS",
         "TORCHINDUCTOR_CACHE_DIR",  # leaked by torch.compile
+        # TensorFlow and TPU related variables
+        "TF2_BEHAVIOR",
+        "TPU_ML_PLATFORM",
+        "TPU_ML_PLATFORM_VERSION",
+        "LD_LIBRARY_PATH",
+        "ENABLE_RUNTIME_UPTIME_TELEMETRY",
     }
     leaked_vars.difference_update(allowlist)
     assert not leaked_vars, f"test is leaking environment variable(s): {set(leaked_vars)}"

tests/tests_pytorch/trainer/flags/test_overfit_batches.py

@@ -170,3 +170,44 @@ def test_distributed_sampler_with_overfit_batches():
     train_sampler = trainer.train_dataloader.sampler
     assert isinstance(train_sampler, DistributedSampler)
     assert train_sampler.shuffle is False
+
+
+def test_overfit_batches_same_batch_for_train_and_val(tmp_path):
+    """Test that when overfit_batches=1, the same batch is used for both training and validation."""
+
+    class TestModel(BoringModel):
+        def __init__(self):
+            super().__init__()
+            self.train_batches = []
+            self.val_batches = []
+
+        def training_step(self, batch, batch_idx):
+            self.train_batches.append(batch)
+            return super().training_step(batch, batch_idx)
+
+        def validation_step(self, batch, batch_idx):
+            self.val_batches.append(batch)
+            return super().validation_step(batch, batch_idx)
+
+    model = TestModel()
+    trainer = Trainer(
+        default_root_dir=tmp_path,
+        max_epochs=2,
+        overfit_batches=1,
+        check_val_every_n_epoch=1,
+        enable_model_summary=False,
+    )
+    trainer.fit(model)
+
+    # Verify that the same batch was used for both training and validation
+    assert len(model.train_batches) > 0
+    assert len(model.val_batches) > 0
+
+    # Compare the actual batch contents
+    train_batch = model.train_batches[0]
+    val_batch = model.val_batches[0]
+
+    # Check if the batches are identical
+    assert torch.equal(train_batch, val_batch), (
+        "Training and validation batches should be identical when overfit_batches=1"
+    )