train.py

#    Copyright 2023 Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li
#
#    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.

from dataclasses import dataclass, field
from typing import Dict, Optional, Sequence

import torch
import transformers
from transformers import Trainer, AddedToken

from visualgpt.dataset import AlpacaDataset, DollyDataset, ShareGPTDataset

IGNORE_INDEX = -100
DEFAULT_PAD_TOKEN = "[PAD]"
DEFAULT_EOS_TOKEN = "</s>"
DEFAULT_BOS_TOKEN = "<s>"
DEFAULT_UNK_TOKEN = "<unk>"
DEFAULT_EOT_TOKEN = "<EOT>"


@dataclass
class ModelArguments:
    model_name_or_path: Optional[str] = field(default="facebook/opt-125m")


@dataclass
class DataArguments:
    databricks_dolly_15k_path: str = field(default=None, metadata={"help": "Path to the training data."})
    sharegpt_data_path: str = field(default=None, metadata={"help": "Path to the training data."})
    evol_instruct_data_path: str = field(default=None, metadata={"help": "Path to the training data."})


@dataclass
class TrainingArguments(transformers.TrainingArguments):
    cache_dir: Optional[str] = field(default=None)
    optim: str = field(default="adamw_torch")
    model_max_length: int = field(
        default=2048,
        metadata={"help": "Maximum sequence length. Sequences will be right padded (and possibly truncated)."},
    )
    gradient_checkpointing: bool = field(default=False)
  
    
def smart_tokenizer_and_embedding_resize(
    special_tokens_dict: Dict,
    other_tokens,
    tokenizer: transformers.PreTrainedTokenizer,
    model: transformers.PreTrainedModel,
):
    """Resize tokenizer and embedding.

    Note: This is the unoptimized version that may make your embedding size not be divisible by 64.
    """
    num_new_tokens = tokenizer.add_special_tokens(special_tokens_dict)
    for new_token in other_tokens:
        num_new_tokens += tokenizer.add_tokens(AddedToken(new_token, normalized=False))

    model.resize_token_embeddings(len(tokenizer))

    if num_new_tokens > 0:
        input_embeddings = model.get_input_embeddings().weight.data
        output_embeddings = model.get_output_embeddings().weight.data

        input_embeddings_avg = input_embeddings[:-num_new_tokens].mean(dim=0, keepdim=True)
        output_embeddings_avg = output_embeddings[:-num_new_tokens].mean(dim=0, keepdim=True)

        input_embeddings[-num_new_tokens:] = input_embeddings_avg
        output_embeddings[-num_new_tokens:] = output_embeddings_avg


@dataclass
class DataCollatorForSupervisedDataset(object):
    """Collate examples for supervised fine-tuning."""

    tokenizer: transformers.PreTrainedTokenizer

    def __call__(self, instances: Sequence[Dict]) -> Dict[str, torch.Tensor]:
        input_ids, labels = tuple([instance[key] for instance in instances] for key in ("input_ids", "labels"))
        input_ids = torch.nn.utils.rnn.pad_sequence(
            input_ids, batch_first=True, padding_value=self.tokenizer.pad_token_id
        )
        labels = torch.nn.utils.rnn.pad_sequence(labels, batch_first=True, padding_value=IGNORE_INDEX)
        return dict(
            input_ids=input_ids,
            labels=labels,
            attention_mask=input_ids.ne(self.tokenizer.pad_token_id),
        )


def make_supervised_data_module(train_dataset, tokenizer: transformers.PreTrainedTokenizer) -> Dict:
    """Make dataset and collator for supervised fine-tuning."""
    data_collator = DataCollatorForSupervisedDataset(tokenizer=tokenizer)
    return dict(train_dataset=train_dataset, eval_dataset=None, data_collator=data_collator)


def train():
    parser = transformers.HfArgumentParser((ModelArguments, DataArguments, TrainingArguments))
    model_args, data_args, training_args = parser.parse_args_into_dataclasses()
    
    model = transformers.AutoModelForCausalLM.from_pretrained(
        model_args.model_name_or_path,
        cache_dir=training_args.cache_dir,
    )

    tokenizer = transformers.AutoTokenizer.from_pretrained(
        model_args.model_name_or_path,
        cache_dir=training_args.cache_dir,
        model_max_length=training_args.model_max_length,
        padding_side="right",
    )

    special_tokens_dict = dict()
    if tokenizer.pad_token is None:
        special_tokens_dict["pad_token"] = DEFAULT_PAD_TOKEN
    if tokenizer.eos_token is None:
        special_tokens_dict["eos_token"] = DEFAULT_EOS_TOKEN
    if tokenizer.bos_token is None:
        special_tokens_dict["bos_token"] = DEFAULT_BOS_TOKEN
    if tokenizer.unk_token is None:
        special_tokens_dict["unk_token"] = DEFAULT_UNK_TOKEN
    
    other_tokens = [DEFAULT_EOT_TOKEN]    
    
    smart_tokenizer_and_embedding_resize(
        special_tokens_dict=special_tokens_dict,
        other_tokens=other_tokens,
        tokenizer=tokenizer,
        model=model,
    )
    
    dataset1 = AlpacaDataset(data_path=data_args.evol_instruct_data_path, tokenizer=tokenizer)
    dataset2 = ShareGPTDataset(data_path=data_args.sharegpt_data_path, tokenizer=tokenizer)
    dataset3 = DollyDataset(data_path=data_args.databricks_dolly_15k_path, tokenizer=tokenizer)
    print(f"dataset1 size:{len(dataset1)}")
    print(f"dataset2 size:{len(dataset2)}")
    print(f"dataset3 size:{len(dataset3)}")
    
    train_dataset = torch.utils.data.ConcatDataset([dataset1, dataset2, dataset3])
    
    data_module = make_supervised_data_module(train_dataset, tokenizer=tokenizer)
    trainer = Trainer(model=model, tokenizer=tokenizer, args=training_args, **data_module)
    trainer.train()
    trainer.save_state()
    trainer.save_model(output_dir=training_args.output_dir)


if __name__ == "__main__":
    train()