TRL¶

This example walks you through how to fine-tune Llama 3.1 with dstack, whether in the cloud or on-prem.

Memory requirements¶

Below are the approximate memory requirements for fine-tuning Llama 3.1.

Model size	Full fine-tuning	LoRA	QLoRA
8B	60GB	16GB	6GB
70B	500GB	160GB	48GB
405B	3.25TB	950GB	250GB

The requirements can be significantly reduced with certain optimizations.

Running on multiple GPUs¶

Below is an example for fine-tuning Llama 3.1 8B using the OpenAssistant/oasst_top1_2023-08-25 dataset:

type: task
name: trl-train

python: "3.10"
# Ensure nvcc is installed (req. for Flash Attention) 
nvcc: true

env:
  - HUGGING_FACE_HUB_TOKEN
  - WANDB_API_KEY
commands:
  - pip install "transformers>=4.43.2"
  - pip install bitsandbytes
  - pip install flash-attn --no-build-isolation
  - pip install peft
  - pip install wandb
  - git clone https://github.com/huggingface/trl
  - cd trl
  - pip install .
  - accelerate launch
    --config_file=examples/accelerate_configs/multi_gpu.yaml
    --num_processes $DSTACK_GPUS_PER_NODE 
    examples/scripts/sft.py
    --model_name meta-llama/Meta-Llama-3.1-8B
    --dataset_name OpenAssistant/oasst_top1_2023-08-25
    --dataset_text_field="text"
    --per_device_train_batch_size 1
    --per_device_eval_batch_size 1
    --gradient_accumulation_steps 4
    --learning_rate 2e-4
    --report_to wandb
    --bf16
    --max_seq_length 1024
    --lora_r 16 --lora_alpha 32
    --lora_target_modules q_proj k_proj v_proj o_proj
    --load_in_4bit
    --use_peft
    --attn_implementation "flash_attention_2"
    --logging_steps=10
    --output_dir models/llama31
    --hub_model_id peterschmidt85/FineLlama-3.1-8B

resources:
gpu:
  # 24GB or more vRAM
  memory: 24GB..
  # One or more GPU
  count: 1..
# Shared memory (for multi-gpu)
shm_size: 24GB

Change the resources property to specify more GPUs.

DeepSpeed¶

For more memory-efficient use of multiple GPUs, consider using DeepSpeed and ZeRO Stage 3.

To do this, use the examples/accelerate_configs/deepspeed_zero3.yaml configuration file instead of examples/accelerate_configs/multi_gpu.yaml.

Running on multiple nodes¶

In case the model doesn't feet into a single GPU, consider running a dstack task on multiple nodes.

type: task
name: trl-train-distrib

# Size of the cluster
nodes: 2

python: "3.10"
# Ensure nvcc is installed (req. for Flash Attention) 
nvcc: true

env:
  - HUGGING_FACE_HUB_TOKEN
  - WANDB_API_KEY
commands:
  - pip install "transformers>=4.43.2"
  - pip install bitsandbytes
  - pip install flash-attn --no-build-isolation
  - pip install peft
  - pip install wandb
  - git clone https://github.com/huggingface/trl
  - cd trl
  - pip install .
  - accelerate launch
    --config_file=examples/accelerate_configs/fsdp_qlora.yaml 
    --main_process_ip=$DSTACK_MASTER_NODE_IP
    --main_process_port=8008
    --machine_rank=$DSTACK_NODE_RANK
    --num_processes=$DSTACK_GPUS_NUM
    --num_machines=$DSTACK_NODES_NUM
      examples/scripts/sft.py
    --model_name meta-llama/Meta-Llama-3.1-8B
    --dataset_name OpenAssistant/oasst_top1_2023-08-25
    --dataset_text_field="text"
    --per_device_train_batch_size 1
    --per_device_eval_batch_size 1
    --gradient_accumulation_steps 4
    --learning_rate 2e-4
    --report_to wandb
    --bf16
    --max_seq_length 1024
    --lora_r 16 --lora_alpha 32
    --lora_target_modules q_proj k_proj v_proj o_proj
    --load_in_4bit
    --use_peft
    --attn_implementation "flash_attention_2"
    --logging_steps=10
    --output_dir models/llama31
    --hub_model_id peterschmidt85/FineLlama-3.1-8B
    --torch_dtype bfloat16
    --use_bnb_nested_quant

resources:
  gpu:
    # 24GB or more vRAM
    memory: 24GB..
    # One or more GPU
    count: 1..
  # Shared memory (for multi-gpu)
  shm_size: 24GB

Fleets¶

By default, dstack run reuses idle instances from one of the existing fleets. If no idle instances meet the requirements, it creates a new fleet using one of the configured backends.

Use fleets configurations to create fleets manually. This reduces startup time for dev environments, tasks, and services, and is very convenient if you want to reuse fleets across runs.

Dev environments¶

Before running a task or service, it's recommended that you first start with a dev environment. Dev environments allow you to run commands interactively.

Source code¶

The source-code of this example can be found in examples/llms/llama31 and examples/fine-tuning/trl .

What's next?¶

Browse the Axolotl and Alignment Handbook examples
Check dev environments, tasks, services, and fleets.