lora_finetune_single_device.py

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the BSD-style license found in the
# LICENSE file in the root directory of this source tree.

import os
import sys
import time
import gc

from functools import partial
from typing import Any, Dict, Optional, Tuple, Union
from warnings import warn

import torch
from omegaconf import DictConfig, ListConfig

from torch import nn
from torch.optim import Optimizer
from torch.utils.data import DataLoader, DistributedSampler
from torchtune import config, modules, training, utils
from torchtune.data import padded_collate_sft
from torchtune.datasets import ConcatDataset
from torchtune.modules.peft import (
    get_adapter_params,
    get_lora_module_names,
    get_merged_lora_ckpt,
    load_dora_magnitudes,
    set_trainable_params,
    validate_missing_and_unexpected_for_lora,
)
from torchtune.recipe_interfaces import FTRecipeInterface
from torchtune.training import DummyProfiler, PROFILER_KEY

from tqdm import tqdm

log = utils.get_logger("DEBUG")


class LoRAFinetuneRecipeSingleDevice(FTRecipeInterface):
    """
    LoRA finetuning recipe for dense transformer-based LLMs such as Llama2. This recipe is optimized
    for single GPU training. Training on CPU is not supported.

    Features:
        - Activation Checkpointing. This can be controlled using the ``activation_checkpointing``
            flag. Activation checkpointing helps reduce the memory footprint since we no longer keep
            activations in memory and instead recompute them during the backward pass. This is especially
            helpful for larger batch sizes when you're memory constrained. But these savings in memory
            come at the cost of training performance. In most cases training can slow-down quite a bit as
            a result of this activation recomputation.

        - Precision. Full fp32 and bf16 training are supported. Precision is controlled using the ``dtype``
            flag. When ``dtype=bf16``, all activations, gradients and optimizer states are in bfloat16. In
            most cases this should halve the memory footprint of full precision (fp32) training, without
            loss in model quality (will depend on the model, training data and other settings). For
            GPUs which do not support bfloat16, we fall back to fp32. Mixed precision training and fp16
            precision are currently not supported.g

        - Gradient Accumulation. You can simulate larger batch sizes by accumulating gradients. This is
            controlled using the ``gradient_accumulation_steps`` flag.

                Total Batch Size = batch_size * gradient accumulation steps.

            For example: with batch_size=1 and gradient_accumulation_steps=32 we get a total batch size of 32.

            Gradient accumulation is especially useful when you are memory constrained. In this case,
            accumulating gradients might give you better training speed than enabling activation
            checkpointing.

        - Lower precision optimizers. This recipe supports lower-precision optimizers from the bitsandbytes
            library (https://huggingface.co/docs/bitsandbytes/main/en/index). We've tested the recipe with
            8-bit AdamW and Paged AdamW.

        - Checkpointing. Model weights are checkpointed both at the end of each epoch and at the end of
            training. Currently we checkpoint both the adapter weights (trainable params only) and the
            complete merged weights (adapter weights added back to the base model). For more details
            please take a look at our LoRA tutorial
            (https://pytorch.org/torchtune/main/tutorials/lora_finetune.html).

            Optimizer State and recipe state (seed, total_epochs, number of epochs run etc) are
            only saved at the end of a given epoch and used in case of resuming training. Resuming
            training is controlled by the ``resume_from_checkpoint`` flag. Mid-epoch checkpointing is
            currently not supported.

            For more details on the checkpointer, please take a look at
            our checkpointer deepdive (https://pytorch.org/torchtune/main/tutorials/checkpointer.html).

        - Logging. Terminal, Disk, WandB and TensorBoard are all supported.

        - Gradient Clipping. Gradient clipping is supported using the ``clip_grad_norm`` flag. By default,
            ``clip_grad_norm`` is set to ``None``. If you only want to log the grad norm, you can set
            ``clip_grad_norm='inf'``.

    For a full list of example configs for this recipe, run ``tune ls`` on the command line. Each config
    has example commands for how to kick-off training.

    Args:
        cfg (DictConfig): OmegaConf object parsed from yaml file

    Raises:
        ValueError: If ``dtype`` is set to fp16.
        RuntimeError: If ``dtype`` is set to bf16 and the hardware does not support bf16.

    """

    def __init__(self, cfg: DictConfig) -> None:

        self._device = utils.get_device(device=cfg.device)
        # Reduced precision logic
        self._dtype = training.get_dtype(cfg.dtype, device=self._device)
        # fp16 precision is explicitly disabled as it is not supported in this
        # recipe (for example, no gradient scaling).
        if self._dtype == torch.float16:
            raise ValueError(
                "fp16 precision is not supported in this recipe. Please use fp32 or bf16."
            )
        # For CUDA devices, check if the HW supports bf16 if bf16 is specified.
        if (
            self._dtype == torch.bfloat16
            and self._device != torch.device("cpu")
            and not torch.cuda.is_bf16_supported()
        ):
            raise RuntimeError("Full bf16 training is not supported on this hardware.")
        # logging attributes
        self._output_dir = cfg.output_dir
        self._log_every_n_steps = cfg.get("log_every_n_steps", 1)
        self._log_peak_memory_stats = cfg.get("log_peak_memory_stats", False)

        # Flag to control whether the output is only an adapter or the full model.
        # If we only output an adapter model, we ignore creating the merged model. This disables intermediate checkpointing.
        #self._adapter_only = cfg.get("adapter_only", False)

        # These are public properties which are updated by the checkpoint loader
        # when ``resume_from_checkpoint`` is `True` or validated in tests
        self.seed = training.set_seed(seed=cfg.seed)
        self.epochs_run = 0
        self.total_epochs = cfg.epochs
        self.max_steps_per_epoch = cfg.max_steps_per_epoch
        self.global_step = 0
        self._resume_from_checkpoint = cfg.resume_from_checkpoint
        self._save_adapter_weights_only = cfg.get("save_adapter_weights_only", False)
        self._gradient_accumulation_steps = cfg.gradient_accumulation_steps
        self._clip_grad_norm = cfg.get("clip_grad_norm", None)

        ### THOMAS EDITS
        # Flag to control whether we save intermediate checkpoints or only the final one.
        self._save_intermediate_checkpoints = cfg.get("save_intermediate_checkpoints", False)

    def load_checkpoint(self, cfg_checkpointer: DictConfig) -> Dict[str, Any]:
        """
        Extract the checkpoint state from file and validate. This includes the
        base model weights. If resume_from_checkpoint is True, this also includes
        the adapter weights and recipe state
        """
        t0 = time.perf_counter()
        self._checkpointer = config.instantiate(
            cfg_checkpointer,
            resume_from_checkpoint=self._resume_from_checkpoint,
        )
        log.info("Loading checkpoint(s) from file…")
        checkpoint_dict = self._checkpointer.load_checkpoint()

        if self._resume_from_checkpoint:
            log.info("Loading adapter checkpoints for resume…")
            if training.ADAPTER_KEY not in checkpoint_dict:
                raise ValueError(
                    "Adapter weights not found. Please ensure a valid adapter checkpoint is provided."
                )
            # _update_recipe_state will throw an exception if the recipe state is not corrctly loaded
            # no need to check here
            self._update_recipe_state(checkpoint_dict)
        t1 = time.perf_counter()
        log.info(f"Checkpoints loaded in {t1-t0:.2f} seconds!")
        return checkpoint_dict

    def _update_recipe_state(self, ckpt_dict: Dict[str, Any]) -> None:
        """
        Updates the recipe state from checkpoint.
        """
        try:
            self.epochs_run = ckpt_dict[training.EPOCHS_KEY]

            # on mismatch, warn the user and prevent the override
            if self.seed != ckpt_dict[training.SEED_KEY]:
                warn(
                    message=(
                        "Config value for seed does not match the checkpoint value, "
                        f"using the checkpoint value: {ckpt_dict[training.SEED_KEY]}"
                    )
                )
                self.seed = ckpt_dict[training.SEED_KEY]
            if self.max_steps_per_epoch != ckpt_dict[training.MAX_STEPS_KEY]:
                warn(
                    message=(
                        "Config value for max_steps_per_epoch does not match the checkpoint value, "
                        f"using the checkpoint value: {ckpt_dict[training.MAX_STEPS_KEY]}"
                    )
                )
                self.max_steps_per_epoch = ckpt_dict[training.MAX_STEPS_KEY]

            # on mismatch, warn the user but allow the override
            if self.total_epochs != ckpt_dict[training.TOTAL_EPOCHS_KEY]:
                warn(
                    message=(
                        "Config value for total_epochs does not match the checkpoint value, "
                        f"using the config value: {self.total_epochs}"
                    )
                )

        except KeyError as e:
            raise KeyError(
                "Checkpoint does not contain the required keys needed for updating recipe state. "
                "Are you sure you passed in the right recipe checkpoint?"
            ) from e

    def setup(self, cfg: DictConfig) -> None:
        """
        Setup the recipe state. This includes recipe state (if resume_from_checkpoint is True),
        model, tokenizer, loss, optimizer, learning rate scheduler, sampler, and dataloader.
        """
        log.info("Starting setup...")
        self._metric_logger = config.instantiate(cfg.metric_logger)

        # log config with parameter override
        self._metric_logger.log_config(cfg)

        self._model_compile = cfg.compile
        checkpoint_dict = self.load_checkpoint(cfg_checkpointer=cfg.checkpointer)

        # set up model
        self._model = self._setup_model(
            cfg_model=cfg.model,
            enable_activation_checkpointing=cfg.enable_activation_checkpointing,
            compile_model=cfg.compile,
            base_model_state_dict=checkpoint_dict[training.MODEL_KEY],
            lora_weights_state_dict=(
                checkpoint_dict[training.ADAPTER_KEY]
                if self._resume_from_checkpoint
                else None
            ),
        )

        self._tokenizer = config.instantiate(cfg.tokenizer)
        log.info("Tokenizer is initialized from file.")

        self._optimizer = self._setup_optimizer(
            cfg_optimizer=cfg.optimizer,
            opt_state_dict=(
                checkpoint_dict[training.OPT_KEY]
                if self._resume_from_checkpoint
                else None
            ),
        )

        # initialize loss
        self._loss_fn = config.instantiate(cfg.loss)
        backend = os.environ.get("TORCH_COMPILE_BACKEND", "inductor")
        if self._loss_fn.__class__.__name__ == "CEWithChunkedOutputLoss":
            # set num_output_chunks for model
            self._model.set_num_output_chunks(self._loss_fn.num_output_chunks)
            if self._model_compile:
                log.info("Compiling loss with torch.compile...")
                # For CEWithChunkedOutputLoss, if we compile the entire class
                # we lose the benefits from the chunked loss.
                # Therefore, we only compile the cross entropy function + upcasting
                self._loss_fn.compute_cross_entropy = torch.compile(
                    self._loss_fn.compute_cross_entropy, backend=backend
                )
        else:
            if self._model_compile:
                log.info("Compiling loss with torch.compile...")
                self._loss_fn = torch.compile(self._loss_fn, backend=backend)
        log.info("Loss is initialized.")

        # Dataloader depends on the tokenizer and loss_fn and should be
        # setup after all of these are setup
        self._sampler, self._dataloader = self._setup_data(
            cfg_dataset=cfg.dataset,
            shuffle=cfg.shuffle,
            batch_size=cfg.batch_size,
        )

        # Load the validation set as well if specified.
        if cfg.get("validation_dataset", None):
            print("Loading validation set…")
            self._validation_sampler, self._validation_dataloader = self._setup_data(
                cfg_dataset=cfg.validation_dataset,
                shuffle=False,
                batch_size=1,
            )
        else:
            self._validation_sampler, self._validation_dataloader = None, None

        # Finally update the recipe state which can only be correctly set after all of the
        # other components have been initialized and updated.

        # Number of training steps in each epoch depends on the number of batches produced
        # by the dataloader and the max_steps_per_epoch param set by the user and is used
        # for logging and tracking training state. This should be computed after the dataloader
        # has been setup
        self._steps_per_epoch = (
            len(self._dataloader) // self._gradient_accumulation_steps
        )
        if (
            self.max_steps_per_epoch is not None
            and self.max_steps_per_epoch < self._steps_per_epoch
        ):
            self._steps_per_epoch = self.max_steps_per_epoch
            self.global_step = self.epochs_run * self._steps_per_epoch

        self._max_validation_steps = cfg.get("max_validation_steps", 0)

        # Learning rate scheduler can only be set up after number of steps
        # has been computed
        self._lr_scheduler = self._setup_lr_scheduler(
            cfg_lr_scheduler=cfg.lr_scheduler,
            num_training_steps=self.total_epochs * self._steps_per_epoch,
            last_epoch=self.global_step - 1,
        )

        # Set up profiler, returns DummyProfiler (nullcontext object with no-op `step` method)
        # if cfg is missing profiler key or if `cfg.profiler.enabled = False
        self._profiler = self._setup_profiler(cfg.get(PROFILER_KEY, None))

        # Used to ignore labels for loss computation
        self.ignore_labels_cache = torch.full(
            (cfg.batch_size, 1), self._loss_fn.ignore_index, device=self._device
        )

    def _setup_profiler(
        self, cfg_profiler: Optional[DictConfig] = None
    ) -> Union[torch.profiler.profile, DummyProfiler]:
        """
        Parses the `profiler` section of top-level `cfg` and sets up profiler

        Args:
            cfg_profiler (Optional[DictConfig]): ``profiler`` section of the top-level ``cfg`` (the main config passed to
                `recipe.main`). Default None.

        Returns:
            profiler: Union[torch.profiler.profile, DummyProfiler] - DummyProfiler is a nullcontext with no-op methods
            for `start`, `stop`, and `step` that can be used in place of `torch.profiler.profile` if profiler is not enabled such
            that the instrumented training loop does not need to be changed profiling is disabled.

        The profiler config can be provided in configs under the `profiler` key with the following layout:

        .. code-block:: yaml
            profiler:
                enabled: bool

                #Output directory of trace artifacts
                output_dir: str

            #`torch.profiler.ProfilerActivity` types to trace
            cpu: bool
            cuda: bool

                #Trace options
                profile_memory: bool
                with_stack: bool
                record_shapes: bool
                with_flops: bool

            # `torch.profiler.schedule` options:
            # wait_steps -> wait, warmup_steps -> warmup, active_steps -> active, num_cycles -> repeat
            wait_steps: int
            warmup_steps: int
            active_steps: int
            num_cycles: int
        """

        # Missing profiler section in config, assume disabled
        if cfg_profiler is None:
            cfg_profiler = DictConfig({"enabled": False})

        # Check that component is included and set correctly
        if cfg_profiler.get("_component_", None) is None:
            cfg_profiler["_component_"] = "torchtune.training.setup_torch_profiler"
        else:
            assert (
                cfg_profiler.get("_component_")
                == "torchtune.training.setup_torch_profiler"
            ), "Only torch profiler supported currently: component must be `torchtune.training.setup_torch_profiler`"

        profiler, profiler_cfg = config.instantiate(cfg_profiler)

        log.info(f" Profiler config after instantiation: {profiler_cfg}")

        self.profiler_profile_memory = profiler_cfg.get("profile_memory", False)
        if profiler_cfg["enabled"]:
            self.profiler_wait_steps = profiler_cfg["wait_steps"]
            self.profiler_warmup_steps = profiler_cfg["warmup_steps"]
            self.profiler_active_steps = profiler_cfg["active_steps"]

        return profiler

    def _setup_model(
        self,
        cfg_model: DictConfig,
        enable_activation_checkpointing: bool,
        compile_model: bool,
        base_model_state_dict: Dict[str, Any],
        lora_weights_state_dict: Optional[Dict[str, Any]] = None,
    ) -> nn.Module:
        log.info("Constructing the model with default param instantiation…")
        t0 = time.perf_counter()

        with training.set_default_dtype(self._dtype), self._device:
            model = config.instantiate(cfg_model)
        t1 = time.perf_counter()
        log.info(f"Model constructed in {t1-t0:.2f} seconds!")

        self._lora_rank = cfg_model.lora_rank
        self._lora_alpha = cfg_model.lora_alpha
        self._lora_attn_modules = list(cfg_model.lora_attn_modules)
        self._apply_lora_to_mlp = cfg_model.apply_lora_to_mlp
        self._apply_lora_to_output = getattr(cfg_model, "apply_lora_to_output", False)
        self.adapter_params = get_adapter_params(model)
        self._is_dora = any(["magnitude" in k for k in self.adapter_params.keys()])
        set_trainable_params(model, self.adapter_params)

        if compile_model:
            log.info("Compiling model layers with torch.compile...")
            backend = os.environ.get("TORCH_COMPILE_BACKEND", "inductor")
            for m in reversed(list(model.modules())):
                if isinstance(m, modules.transformer.TransformerSelfAttentionLayer):
                    m.compile(backend=backend)

        if enable_activation_checkpointing:
            log.info("Loading activation checkpoints…")
            training.set_activation_checkpointing(
                model, auto_wrap_policy={modules.TransformerSelfAttentionLayer}
            )

        log.info("Loading base state_dicts…")
        t0 = time.perf_counter()
        base_missing, base_unexpected = model.load_state_dict(
            base_model_state_dict, strict=False
        )
        t1 = time.perf_counter()
        log.info(f"Base state dicts loaded in {t1-t0:.2f} seconds!")

        # This is for any adapters that need to be initialized after base weights
        # have been loaded (e.g. DoRA).
        if self._is_dora:
            load_dora_magnitudes(model)

        if lora_weights_state_dict:
            log.info("Loading LoRA state dict(s)…")
            t0 = time.perf_counter()
            lora_missing, lora_unexpected = model.load_state_dict(
                lora_weights_state_dict, strict=False
            )
            t1 = time.perf_counter()
            log.info(f"LoRA state dicts loaded in {t1-t0:.2f} seconds!")
        else:
            lora_missing, lora_unexpected = None, None
        
        #log.info("Validate missing and unexpected for LoRA…")
        validate_missing_and_unexpected_for_lora(
            lora_attn_modules=self._lora_attn_modules,
            apply_lora_to_mlp=self._apply_lora_to_mlp,
            apply_lora_to_output=self._apply_lora_to_output,
            base_missing=base_missing,
            base_unexpected=base_unexpected,
            lora_missing=lora_missing,
            lora_unexpected=lora_unexpected,
        )
        # Validate model adapter params were loaded in with the expected dtype
        # TODO (rohan-varma): Further validation to ensure the appropriate base params
        # are NF4 vs bf16 based on the quantization config.
        training.validate_expected_param_dtype(
            self.adapter_params.items(), dtype=self._dtype
        )

        log.info(f"Model is initialized with precision {self._dtype}.")

        if self._device.type == "cuda":
            memory_stats = training.get_memory_stats(device=self._device)
            training.log_memory_stats(memory_stats)
        return model

    def _setup_optimizer(
        self, cfg_optimizer: DictConfig, opt_state_dict: Optional[Dict[str, Any]] = None
    ) -> Optimizer:
        optimizer = config.instantiate(cfg_optimizer, self._model.parameters())
        if opt_state_dict:
            optimizer.load_state_dict(opt_state_dict)

        log.info("Optimizer and loss are initialized.")
        return optimizer

    def _setup_lr_scheduler(
        self,
        cfg_lr_scheduler: DictConfig,
        num_training_steps: int,
        last_epoch: int,
    ) -> Optimizer:
        lr_scheduler = config.instantiate(
            cfg_lr_scheduler,
            self._optimizer,
            num_training_steps=num_training_steps,
            last_epoch=last_epoch,
        )

        log.info("Learning rate scheduler is initialized.")
        return lr_scheduler

    def _setup_data(
        self,
        cfg_dataset: DictConfig,
        shuffle: bool,
        batch_size: int,
    ) -> Tuple[DistributedSampler, DataLoader]:
        """
        All data related setup happens here. Currently this recipe only supports
        Map-style Datasets which fit into memory and an option for random shuffling.
        Samplers, iterable datasets, and streaming datasets are not supported.
        """
        if isinstance(cfg_dataset, ListConfig):
            datasets = [
                config.instantiate(single_cfg_dataset, self._tokenizer)
                for single_cfg_dataset in cfg_dataset
            ]
            ds = ConcatDataset(datasets=datasets)
            packed = False
        else:
            ds = config.instantiate(cfg_dataset, self._tokenizer)
            packed = cfg_dataset.get("packed", False)

        log.info(f"Loaded dataset with {len(ds)} samples.")

        sampler = DistributedSampler(
            ds,
            num_replicas=1,
            rank=0,
            shuffle=shuffle,
            seed=0,
        )
        dataloader = DataLoader(
            dataset=ds,
            sampler=sampler,
            batch_size=batch_size,
            collate_fn=(
                partial(
                    padded_collate_sft,
                    padding_idx=self._tokenizer.pad_id,
                    ignore_idx=self._loss_fn.ignore_index,
                )
                if not packed
                else None
            ),
        )

        log.info("Dataset and Sampler are initialized.")

        return sampler, dataloader

    def save_checkpoint(self, epoch: int) -> None:
        """
        Checkpoint the state of the recipe. The constructed checkpoint state dict
        contains the following information:
        - Merged weights with key MODEL_KEY
        - Adapter weights with key ADAPTER_KEY
        - Relevant recipe state if training is not complete
        - If the `self._save_adapter_weights_only` option is True, the checkpointer will save only the adapter weights

        To correctly resume from training, the adapter weights and recipe state must be provided along with the base model weights.
        """
        intermediate_checkpoint = epoch + 1 < self.total_epochs

        ### THOMAS EDITS
        # If we're not saving intermediate checkpoints, skip the checkpointing
        if intermediate_checkpoint and not self._save_intermediate_checkpoints:
            log.info("Skipping intermediate checkpointing.")
            return None

        ckpt_dict = {}
        
        # if training is in-progress, checkpoint the optimizer state as well
        if intermediate_checkpoint:
            log.info("Saving optimizer state as well because training is not yet complete.")
            ckpt_dict.update(
                {
                    training.OPT_KEY: self._optimizer.state_dict(),
                    training.SEED_KEY: self.seed,
                    training.EPOCHS_KEY: self.epochs_run,
                    training.TOTAL_EPOCHS_KEY: self.total_epochs,
                    training.MAX_STEPS_KEY: self.max_steps_per_epoch,
                }
            )
        else:
            log.info("Saving final checkpoint...")

        # Move to CPU to avoid a copy on GPU
        log.debug("Moving model to CPU...")
        state_dict = {k: v.cpu() for k, v in self._model.state_dict().items()}

        # Construct the full state dict with LoRA weights merged into base LLM weights
        #log.debug("Constructing the full state dict with LoRA weights merged into base LLM weights.")
        if self._save_adapter_weights_only:
            log.info("We're skipping the construction of the merged dict to save time.")
            merged_state_dict = state_dict
        else:
            log.info("Creating merged state dict…")
            merged_state_dict = get_merged_lora_ckpt(
                state_dict,
                rank=self._lora_rank,
                alpha=self._lora_alpha,
            )
        ckpt_dict.update({training.MODEL_KEY: merged_state_dict})

        # Construct the adapter weights
        log.info("Constructing the LoRA adapter state dict.")
        adapter_key_filter = lambda x: x in self.adapter_params
        adapter_state_dict = {
            k: v for k, v in self._model.state_dict().items() if adapter_key_filter(k)
        }
        ckpt_dict.update({training.ADAPTER_KEY: adapter_state_dict})
        adapter_config = {
            "r": self._lora_rank,
            "lora_alpha": self._lora_alpha,
            "target_modules": get_lora_module_names(
                self._lora_attn_modules,
                self._apply_lora_to_mlp,
                self._apply_lora_to_output,
            ),
            "peft_type": "LORA",
        }
        ckpt_dict.update({training.ADAPTER_CONFIG: adapter_config})

        self._checkpointer.save_checkpoint(
            ckpt_dict,
            epoch=epoch,
            intermediate_checkpoint=intermediate_checkpoint,
            adapter_only=self._save_adapter_weights_only,
        )
        log.info("Checkpoint saved!")

    def _loss_step(self, batch: Dict[str, torch.Tensor]) -> torch.Tensor:
        # Both are shape [b, s]
        tokens, labels = batch["tokens"], batch["labels"]

        # Get the attention mask and position ids from the dataset if they
        # exist. Currently, only sample packing in PackedDataset returns these
        mask = batch.get("mask", None)  # shape [b, s, s]
        input_pos = batch.get("input_pos", None)  # shape [b, s]

        # run model
        logits = self._model(tokens, mask=mask, input_pos=input_pos)

        # Shift labels to compute loss
        # equivalent to doing labels[..., 1:] and logits[..., :-1, :]
        # But this way we dont need to slice the logits. We just add an ignore index to labels.
        labels = torch.hstack(
            (labels[..., 1:], self.ignore_labels_cache[: labels.shape[0]])
        )
        if not isinstance(logits, list):
            labels = labels.reshape(-1)
            logits = logits.reshape(-1, logits.size(-1))

        # Compute loss
        loss = self._loss_fn(logits, labels)

        # free logits otherwise it peaks backward memory
        del logits

        return loss

    def train(self) -> None:
        """
        The core training loop. Supports training on subsets of the dataset using the
        ``max_steps_per_epoch``.
        """
        if self._model_compile:
            log.info(
                "NOTE: torch.compile is enabled and model is compiled in first forward. Expect a relatively slow first iteration."
            )
        
        ### EPOCH LOOP
        log.info(f"Starting training from epoch {self.epochs_run+1} out of {self.total_epochs}...")
        for curr_epoch in range(self.epochs_run, self.total_epochs):
            # Update the sampler to ensure data is correctly shuffled across epochs
            # in case shuffle is True
            self._sampler.set_epoch(curr_epoch)
            self._model.train()

            # Initialize tokens count and running loss (for grad accumulation)
            running_loss = 0
            num_tokens = 0
            t0 = time.perf_counter()

            # Clear memory from previous validation run
            gc.collect()
            torch.cuda.empty_cache()

            ### GLOBAL STEP LOOP
            pbar = tqdm(total=self._steps_per_epoch)
            for idx, batch in enumerate(self._dataloader):
                if (
                    self.max_steps_per_epoch is not None
                    and (idx // self._gradient_accumulation_steps)
                    == self.max_steps_per_epoch
                ):
                    break

                batch = {k: v.to(self._device) for k, v in batch.items()}
                num_tokens += batch["tokens"].numel()

                loss = self._loss_step(batch)
                loss = loss / self._gradient_accumulation_steps
                running_loss += loss
                loss.backward()

                # Step with optimizer
                if (idx + 1) % self._gradient_accumulation_steps == 0:
                    if self._clip_grad_norm is not None:
                        grad_norm = torch.nn.utils.clip_grad_norm_(
                            self._model.parameters(),
                            max_norm=float(self._clip_grad_norm),
                        )
                    self._optimizer.step()
                    self._optimizer.zero_grad(set_to_none=True)
                    self._lr_scheduler.step()

                    # Update global step
                    self.global_step += 1

                    loss_to_log = running_loss.item()
                    pbar.update(1)
                    pbar.set_description(
                        f"{curr_epoch + 1}|{self.global_step}|Loss: {loss_to_log}"
                    )

                    # Log per-step metrics
                    if self.global_step % self._log_every_n_steps == 0:
                        time_per_step = time.perf_counter() - t0
                        log_dict = {
                            "loss": loss_to_log,
                            "lr": self._optimizer.param_groups[0]["lr"],
                            "tokens_per_second_per_gpu": num_tokens / time_per_step,
                        }
                        if (
                            self._device.type == "cuda"
                            and self._log_peak_memory_stats
                        ):
                            log_dict.update(
                                training.get_memory_stats(device=self._device)
                            )
                        if self._clip_grad_norm is not None:
                            log_dict.update({"grad_norm": grad_norm})
                        self._metric_logger.log_dict(
                            log_dict,
                            step=self.global_step,
                        )

                    # Reset running stats for the next global step
                    running_loss = 0
                    num_tokens = 0
                    t0 = time.perf_counter()

            ### VALIDATION LOOP
            # Run the validation after each epoch
            if self._validation_dataloader:
                validation_steps = min(self._max_validation_steps, len(self._validation_dataloader))

                running_loss = 0
                t0 = time.perf_counter()

                self._model.eval()
                gc.collect()
                torch.cuda.empty_cache()
                
                pbar = tqdm(enumerate(self._validation_dataloader), total=validation_steps, leave=False)
                for idx, batch in pbar:
                    if idx == validation_steps:
                        break
                    
                    with torch.no_grad():
                        batch = {k: v.to(self._device) for k, v in batch.items()}
                        loss = self._loss_step(batch)
                    
                    running_loss += loss.item()
                    pbar.set_description(
                        f"Validation|{self.global_step}|Loss: {loss:.6f}"
                    )
                
                self._metric_logger.log_dict(
                    {'val_loss': running_loss / validation_steps},
                    step=self.global_step,
                )
                log.info(
                    f"Validation loss: {running_loss / validation_steps:.4f} | Epoch: {curr_epoch+1} | Samples: {validation_steps} | Time: {time.perf_counter() - t0:.2f}s"
                )
            
            self.save_checkpoint(epoch=curr_epoch)
            self.epochs_run += 1
            
            ### END OF EPOCH
        
        # END OF TRAINING
        log.info("Training loop done!")

    def cleanup(self) -> None:
        self._metric_logger.close()


@config.parse
def recipe_main(cfg: DictConfig) -> None:
    """
    Entry point for the recipe.

    Configurable parameters are read in the following order:
        - Parameters specified in config (see available configs through ``tune ls``)
        - Overwritten by arguments from the command-line
    """
    config.log_config(recipe_name="LoRAFinetuneRecipeSingleDevice", cfg=cfg)
    recipe = LoRAFinetuneRecipeSingleDevice(cfg=cfg)
    recipe.setup(cfg=cfg)
    recipe.train()
    recipe.cleanup()    # Closes the metric logger


if __name__ == "__main__":
    sys.exit(recipe_main())