base_model: JackFram/llama-68m

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: arcee-ai/distilabel-intel-orca-dpo-pairs-binarized
    type: chatml.ultra
    split: train
dataset_prepared_path: last_run_prepared
val_set_size: 0.1
output_dir: ./outputs/out

sequence_len: 2048
pad_to_sequence_len: true

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 1
num_epochs: 1
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 2e-5

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
flash_attention: true

rl: dpo
dpo_use_weighting: true

warmup_steps: 10
evals_per_epoch: 2
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>