base_model: deepseek-ai/DeepSeek-R1-Distill-Llama-70B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
trust_remote_code: true

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: ugaoo/transformed_short_answer_dataset_prompt
    type: alpaca
val_set_size: 0
output_dir: ./out/transformed_short_answer_dataset_prompt_deepseek70bllama

sequence_len: 4000
sample_packing: true
pad_to_sequence_len: true

adapter: qlora
lora_r: 256
lora_alpha: 512
lora_dropout: 0.05
lora_target_linear: true
lora_target_modules:
  - q_proj
  - k_proj
  - v_proj
  - o_proj
  - up_proj
  - down_proj
  - gate_proj
lora_modules_to_save:
  - embed_tokens
  - lm_head
 
wandb_project: cosmosearch
wandb_entity:
wandb_watch: 
wandb_name: transformed_short_answer_dataset_prompt_deepseek70bllama
wandb_log_model:

gradient_accumulation_steps: 3
micro_batch_size: 4
num_epochs: 6
optimizer: adamw_torch
lr_scheduler: cosine
learning_rate: 5e-6

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

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

warmup_steps: 100
evals_per_epoch: 6
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
save_total_limit: 6
special_tokens:
   pad_token: <｜end▁of▁sentence｜>