recipes/constitutional-ai/dpo/config_anthropic.yaml

# Model arguments
model_name_or_path: alignment-handbook/mistral-7b-sft-constitutional-ai
torch_dtype: null

# Data training arguments
# For definitions, see: src/h4/training/config.py
dataset_mixer:
  HuggingFaceH4/ultrafeedback_binarized: 1.0
  HuggingFaceH4/cai-conversation-harmless: 1.0
dataset_splits:
- train_prefs
- test_prefs
preprocessing_num_workers: 12

# DPOTrainer arguments
bf16: true
beta: 0.1
do_eval: true
do_train: true
eval_strategy: steps
eval_steps: 1000
gradient_accumulation_steps: 1
gradient_checkpointing: true
hub_model_id: mistral-7b-dpo-constitutional-ai
learning_rate: 5.0e-7
log_level: info
logging_steps: 10
lr_scheduler_type: linear
max_length: 1024
max_prompt_length: 512
num_train_epochs: 3
optim: rmsprop
output_dir: data/mistral-7b-dpo-constitutional-ai
per_device_train_batch_size: 2
per_device_eval_batch_size: 8
push_to_hub: true
save_strategy: "steps"
save_steps: 100
save_total_limit: 1
seed: 42
warmup_ratio: 0.1