13B_qlora_single_device.yaml

# Config for single device QLoRA with lora_finetune_single_device.py
# using a Llama2 13B model
#
# This config assumes that you've run the following command before launching
# this run:
#   tune download meta-llama/Llama-2-13b-hf --output-dir /tmp/Llama-2-13b-hf --hf-token <HF_TOKEN>
#
# To launch on a single device, run the following command from root:
#   tune run lora_finetune_single_device --config llama2/13B_qlora_single_device
#
# You can add specific overrides through the command line. For example
# to override the checkpointer directory while launching training
# you can run:
#   tune run lora_finetune_single_device --config 13_qlora_single_device checkpointer.checkpoint_dir=<YOUR_CHECKPOINT_DIR>
#
# This config works only for training on single device.

# Model Arguments
model:
  _component_: torchtune.models.llama2.qlora_llama2_13b
  lora_attn_modules: ['q_proj', 'v_proj', 'k_proj', 'output_proj']
  apply_lora_to_mlp: True
  apply_lora_to_output: False
  lora_rank: 8
  lora_alpha: 16
  lora_dropout: 0.0

tokenizer:
  _component_: torchtune.models.llama2.llama2_tokenizer
  path: /tmp/Llama-2-13b-hf/tokenizer.model
  max_seq_len: null

checkpointer:
  _component_: torchtune.training.FullModelHFCheckpointer
  checkpoint_dir: /tmp/Llama-2-13b-hf/
  checkpoint_files: [
    pytorch_model-00001-of-00003.bin,
    pytorch_model-00002-of-00003.bin,
    pytorch_model-00003-of-00003.bin
  ]
  adapter_checkpoint: null
  recipe_checkpoint: null
  output_dir: /tmp/Llama-2-13b-hf/
  model_type: LLAMA2
resume_from_checkpoint: False
save_adapter_weights_only: False

# Dataset and Sampler
dataset:
  packed: False # Set to true for great speed ups
  _component_: torchtune.datasets.alpaca_cleaned_dataset
seed: null
shuffle: True
batch_size: 2

# Optimizer and Scheduler
optimizer:
  _component_: torch.optim.AdamW
  fused: True
  weight_decay: 0.01
  lr: 3e-4
lr_scheduler:
  _component_: torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup
  num_warmup_steps: 100

loss:
  _component_: torchtune.modules.loss.CEWithChunkedOutputLoss

# Training
epochs: 1
max_steps_per_epoch: null
gradient_accumulation_steps: 16
compile: False

# Logging
output_dir: /tmp/qlora_finetune_output/
metric_logger:
  _component_: torchtune.training.metric_logging.DiskLogger
  log_dir: ${output_dir}
log_every_n_steps: 1
log_peak_memory_stats: True

# Environment
device: cuda
dtype: bf16

enable_activation_checkpointing: True
enable_activation_offloading: False

# Show case the usage of pytorch profiler
# Set enabled to False as it's only needed for debugging training
profiler:
  _component_: torchtune.training.setup_torch_profiler
  enabled: False

  #Output directory of trace artifacts
  output_dir: ${output_dir}/profiling_outputs

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

  #trace options passed to `torch.profiler.profile`
  profile_memory: False
  with_stack: False
  record_shapes: True
  with_flops: False

  # `torch.profiler.schedule` options:
  # wait_steps -> wait, warmup_steps -> warmup, active_steps -> active, num_cycles -> repeat
  wait_steps: 5
  warmup_steps: 5
  active_steps: 2
  num_cycles: 1