Out of memory issue in Lllama3.1 8B finetuning

Question

I am trying to finetune Llama3.1 8B. I am using 4 A10G gpu with 24 GB each. I have model weight present on the local file in base_model = '<model_weight_folder>'

from accelerate import PartialState
device_string = PartialState().process_index

torch_dtype = torch.float16
attn_implementation = "eager"  

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch_dtype,
    bnb_4bit_use_double_quant=True
)

# Load model
model = AutoModelForCausalLM.from_pretrained(
    base_model,
    quantization_config=bnb_config,
    # device_map={'':torch.cuda.current_device()},
    # device_map="auto",
    device_map={'':device_string},
    attn_implementation=attn_implementation
)




# LoRA config
peft_config = LoraConfig(
    r=16,
    lora_alpha=32,
    lora_dropout=0.05,
    bias="none",
    task_type="CAUSAL_LM",
    target_modules=['up_proj', 'down_proj', 'gate_proj', 'k_proj', 'q_proj', 'v_proj', 'o_proj']
)
model = get_peft_model(model, peft_config)




training_arguments = TrainingArguments(
    output_dir=new_model,
    per_device_train_batch_size=1,
    per_device_eval_batch_size=1,
    gradient_accumulation_steps=2,
    # gradient_checkpointing_kwargs={'use_reentrant':False},
    optim="paged_adamw_32bit",
    num_train_epochs=5,
    evaluation_strategy="steps",
    eval_steps=0.2,
    logging_steps=1,
    warmup_steps=10,
    logging_strategy="steps",
    learning_rate=2e-4,
    fp16=False,
    bf16=False,
    group_by_length=True,
    report_to="tensorboard"
)

trainer = SFTTrainer(
    model=model,
    train_dataset=dataset["train"],
    eval_dataset=dataset["test"],
    peft_config=peft_config,
    # max_seq_length=512,
    # dataset_text_field="text",
    tokenizer=tokenizer,
    args=training_arguments,
    # packing= False,
)
trainer.train()

I am launching it using

python -m torch.distributed.launch trainer.py

However it is resulting into out of memory issue while loading the base model.

Can anyone suggest if I can use multi gpu for finetuning?. If yes, what config change is needed.