configs/dpo.yml

base_model: teknium/OpenHermes-2.5-Mistral-7B
model_type: MistralForCausalLM
tokenizer_type: LlamaTokenizer
is_mistral_derived_model: true

load_in_8bit: false
load_in_4bit: true
strict: false

rl: true
datasets:
  - path: Intel/orca_dpo_pairs
    split: train
    type: intel_apply_chatml
  - path: argilla/ultrafeedback-binarized-preferences
    split: train
    type: argilla_apply_chatml
dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./dpopenhermes-alpha-v1/

adapter: qlora
lora_model_dir:

sequence_len: 2048
sample_packing: false
pad_to_sequence_len: false

lora_r: 64
lora_alpha: 32
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

wandb_project: openhermes-dpo
wandb_entity: oaaic
wandb_watch:
wandb_run_id:
wandb_log_model:

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

train_on_inputs: false
group_by_length: false
bf16: true
fp16: false
tf32: true

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 10
eval_steps:
eval_table_size:
eval_table_max_new_tokens: 128
save_steps: 239
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
save_safetensors: true