axolotl version: 0.4.0

base_model: beomi/OPEN-SOLAR-KO-10.7B

load_in_8bit: false
load_in_4bit: false
strict: false

rl: dpo
datasets:
  - path: datasets/dposet/dpodatav2.jsonl
    ds_type: json
    data_files:
      - datasets/dposet/dpodatav2.jsonl
    split: train

dataset_prepared_path:
val_set_size: 0.0
output_dir: ./beomidpo-out-v2

adapter: lora
lora_model_dir:

sequence_len: 2048
sample_packing: false
pad_to_sequence_len: false

lora_r: 8
lora_alpha: 32
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
lora_target_modules:
  - q_proj
  - v_proj
  - k_proj
  - o_proj
gradient_accumulation_steps: 1
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: false
fp16: true
tf32: false

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

warmup_steps: 10
save_steps: 100
save_total_limit: 3
debug:
deepspeed: deepspeed_configs/zero2.json
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
save_safetensors: false