# Config for multi-device LoRA finetuning in lora_finetune_distributed.py
# using a Llama3 8B model
#
# This config assumes that you've run the following command before launching
# this run:
#   tune download meta-llama/Meta-Llama-3-8B --output-dir /tmp/Meta-Llama-3-8B --hf-token <HF_TOKEN>
#
# To launch on 2 devices, run the following command from root:
#   tune run --nproc_per_node 2 lora_finetune_distributed --config llama3/8B_lora
#
# You can add specific overrides through the command line. For example
# to override the checkpointer directory while launching training
# you can run:
#   tune run --nproc_per_node 2 lora_finetune_distributed --config llama3/8B_lora checkpointer.checkpoint_dir=<YOUR_CHECKPOINT_DIR>
#
# This config works best when the model is being fine-tuned on 2+ GPUs.
# For single device LoRA finetuning please use 8B_lora_single_device.yaml
# or 8B_qlora_single_device.yaml

# Tokenizer
tokenizer:
  _component_: torchtune.models.llama3.llama3_tokenizer
  path: ./model/original/tokenizer.model

# Model Arguments
model:
  _component_: torchtune.models.llama3.lora_llama3_8b
  lora_attn_modules: ['q_proj', 'v_proj']
  apply_lora_to_mlp: False
  apply_lora_to_output: False
  lora_rank: 8
  lora_alpha: 16

checkpointer:
  _component_: torchtune.utils.FullModelMetaCheckpointer
  checkpoint_dir: ./model/original/
  checkpoint_files: [
    consolidated.00.pth
  ]
  recipe_checkpoint: null
  output_dir: ./finetuned_model/
  model_type: LLAMA3
resume_from_checkpoint: False

# Dataset and Sampler
# InstructDataset(
#        tokenizer=tokenizer,
#        source=source,
#        template=GrammarErrorCorrectionTemplate,
#        column_map={"sentence": "input"},
#        train_on_input=train_on_input,
#        split="train",
#    )
dataset:
  _component_: torchtune.datasets.instruct_dataset 
  source: grammarly/coedit
  template: GrammarErrorCorrectionTemplate
  column_map: {"sentence": "src", "output": "tgt"}
  train_on_input: False
  split: train
seed: 123
shuffle: True
batch_size: 4

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

loss:
  _component_: torch.nn.CrossEntropyLoss

# Training
epochs: 2
max_steps_per_epoch: null
gradient_accumulation_steps: 32

# Logging
output_dir: ./lora_finetune_output
metric_logger:
  _component_: torchtune.utils.metric_logging.WandBLogger
  project: torchtune
  group: llama3-grammarly
log_every_n_steps: null

# Environment
device: cuda
dtype: bf16
enable_activation_checkpointing: False