sft_vlm.py

# Copyright 2024 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
pip install deepspeed
pip install trl
pip install pillow
pip install transformers>=4.45.1

# Tested on 8x H100 GPUs
accelerate launch --config_file=deepspeed_zero3.yaml scripts/sft_vlm.py \
    --model_name_or_path meta-llama/Llama-3.2-11B-Vision-Instruct \
    --dataset_name HuggingFaceH4/llava-instruct-mix-vsft \
    --per_device_train_batch_size 8 \
    --gradient_accumulation_steps 8 \
    --output_dir llama-3.2-11b-vision-sft \
    --bf16 \
    --torch_dtype bfloat16 \
    --gradient_checkpointing
"""

from trl.commands.cli_utils import SFTScriptArguments, TrlParser
import torch
from accelerate import Accelerator
from datasets import load_dataset

from transformers import AutoModelForVision2Seq, AutoProcessor, LlavaForConditionalGeneration

from trl import (
    ModelConfig,
    SFTConfig,
    SFTTrainer,
    get_peft_config,
    get_quantization_config,
    get_kbit_device_map,
)


if __name__ == "__main__":
    parser = TrlParser((SFTScriptArguments, SFTConfig, ModelConfig))
    sft_script_args, training_args, model_config = parser.parse_args_and_config()
    training_args.gradient_checkpointing_kwargs = dict(use_reentrant=False)
    training_args.dataset_text_field = ""  # need a dummy field
    training_args.remove_unused_columns = False
    training_args.dataset_kwargs = {"skip_prepare_dataset": True}

    ################
    # Model, Tokenizer & Processor
    ################
    torch_dtype = (
        model_config.torch_dtype
        if model_config.torch_dtype in ["auto", None]
        else getattr(torch, model_config.torch_dtype)
    )
    quantization_config = get_quantization_config(model_config)
    model_kwargs = dict(
        revision=model_config.model_revision,
        attn_implementation=model_config.attn_implementation,
        torch_dtype=torch_dtype,
        device_map=get_kbit_device_map() if quantization_config is not None else None,
        quantization_config=quantization_config,
    )
    processor = AutoProcessor.from_pretrained(
        model_config.model_name_or_path, trust_remote_code=model_config.trust_remote_code
    )

    model = AutoModelForVision2Seq.from_pretrained(
        model_config.model_name_or_path, trust_remote_code=model_config.trust_remote_code, **model_kwargs
    )

    ################
    # Create a data collator to encode text and image pairs
    ################
    def collate_fn(examples):
        # Get the texts and images, and apply the chat template
        texts = [processor.apply_chat_template(example["messages"], tokenize=False) for example in examples]
        images = [example["images"] for example in examples]
        if isinstance(model, LlavaForConditionalGeneration):
            # LLava1.5 does not support multiple images
            images = [image[0] for image in images]

        # Tokenize the texts and process the images
        batch = processor(text=texts, images=images, return_tensors="pt", padding=True)

        # The labels are the input_ids, and we mask the padding tokens in the loss computation
        labels = batch["input_ids"].clone()
        labels[labels == processor.tokenizer.pad_token_id] = -100  #
        # Ignore the image token index in the loss computation (model specific)
        image_token_id = processor.tokenizer.convert_tokens_to_ids(processor.image_token)
        labels[labels == image_token_id] = -100
        batch["labels"] = labels

        return batch

    ################
    # Dataset
    ################
    dataset = load_dataset(sft_script_args.dataset_name)

    ################
    # Training
    ################
    trainer = SFTTrainer(
        model=model,
        args=training_args,
        data_collator=collate_fn,
        train_dataset=dataset[sft_script_args.dataset_train_split],
        eval_dataset=dataset[sft_script_args.dataset_test_split],
        tokenizer=processor.tokenizer,
        peft_config=get_peft_config(model_config),
    )

    trainer.train()

    trainer.save_model(training_args.output_dir)
    trainer.push_to_hub()
    if Accelerator().is_main_process:
        processor.push_to_hub(training_args.hub_model_id)