llava/model/preprocessor.py

import copy
import transformers
import tokenizers
import torch
from typing import Dict, Optional, Sequence, List
from packaging import version

from llava.mm_utils import tokenizer_image_token
from llava.train.arguments import ModelArguments, TrainingArguments, DataArguments
from llava.constants import IGNORE_INDEX, MM_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN, DEFAULT_VIDEO_TOKEN, DEFAULT_VIDEO_START_TOKEN, DEFAULT_VIDEO_END_TOKEN
from llava import conversation as conversation_lib

IS_TOKENIZER_GREATER_THAN_0_14 = version.parse(tokenizers.__version__) >= version.parse('0.14')

def _tokenize_fn(strings: Sequence[str],
                 tokenizer: transformers.PreTrainedTokenizer) -> Dict:
    """Tokenize a list of strings."""
    tokenized_list = [
        tokenizer(
            text,
            return_tensors="pt",
            padding="longest",
            max_length=tokenizer.model_max_length,
            truncation=True,
        ) for text in strings
    ]
    input_ids = labels = [
        tokenized.input_ids[0] for tokenized in tokenized_list
    ]
    input_ids_lens = labels_lens = [
        tokenized.input_ids.ne(tokenizer.pad_token_id).sum().item()
        for tokenized in tokenized_list
    ]
    return dict(
        input_ids=input_ids,
        labels=labels,
        input_ids_lens=input_ids_lens,
        labels_lens=labels_lens,
    )


def _mask_targets(target, tokenized_lens, speakers):
    # cur_idx = 0
    cur_idx = tokenized_lens[0]
    tokenized_lens = tokenized_lens[1:]
    target[:cur_idx] = IGNORE_INDEX
    # TODO: why +2 ?
    for tokenized_len, speaker in zip(tokenized_lens, speakers):
        if speaker == "human":
            target[cur_idx+2:cur_idx + tokenized_len] = IGNORE_INDEX
        cur_idx += tokenized_len


def _add_speaker_and_signal(header, source, get_conversation=True):
    """Add speaker and start/end signal on each round."""
    BEGIN_SIGNAL = "### "
    END_SIGNAL = "\n"
    conversation = header
    for sentence in source:
        from_str = sentence["from"]
        if from_str.lower() == "human":
            from_str = conversation_lib.default_conversation.roles[0]
        elif from_str.lower() == "gpt":
            from_str = conversation_lib.default_conversation.roles[1]
        else:
            from_str = 'unknown'
        sentence["value"] = (BEGIN_SIGNAL + from_str + ": " +
                             sentence["value"] + END_SIGNAL)
        if get_conversation:
            conversation += sentence["value"]
    conversation += BEGIN_SIGNAL
    return conversation

def preprocess_multimodal(
    sources: Sequence[str],
    data_args: DataArguments
) -> Dict:
    is_multimodal = data_args.is_multimodal
    if not is_multimodal:
        return sources

    for source in sources:
        for sentence in source:

            if DEFAULT_VIDEO_TOKEN in sentence['value']:
                # sentence['value'] = sentence['value'].replace(DEFAULT_IMAGE_TOKEN, '').strip()
                # sentence['value'] = DEFAULT_IMAGE_TOKEN + '\n' + sentence['value']
                sentence['value'] = sentence['value'].strip()
                if "mmtag" in conversation_lib.default_conversation.version:
                    raise NotImplementedError
                    # sentence['value'] = sentence['value'].replace(DEFAULT_IMAGE_TOKEN, '<Image>' + DEFAULT_IMAGE_TOKEN + '</Image>')
                replace_token = DEFAULT_VIDEO_TOKEN
                if data_args.mm_use_start_end:
                    replace_token = DEFAULT_VIDEO_START_TOKEN + replace_token + DEFAULT_VIDEO_END_TOKEN
                sentence["value"] = sentence["value"].replace(DEFAULT_VIDEO_TOKEN, replace_token)
            

            if DEFAULT_IMAGE_TOKEN in sentence['value']:
                # sentence['value'] = sentence['value'].replace(DEFAULT_IMAGE_TOKEN, '').strip()
                # sentence['value'] = DEFAULT_IMAGE_TOKEN + '\n' + sentence['value']
                sentence['value'] = sentence['value'].strip()
                if "mmtag" in conversation_lib.default_conversation.version:
                    sentence['value'] = sentence['value'].replace(DEFAULT_IMAGE_TOKEN, '<Image>' + DEFAULT_IMAGE_TOKEN + '</Image>')
                replace_token = DEFAULT_IMAGE_TOKEN
                if data_args.mm_use_start_end:
                    replace_token = DEFAULT_IM_START_TOKEN + replace_token + DEFAULT_IM_END_TOKEN
                sentence["value"] = sentence["value"].replace(DEFAULT_IMAGE_TOKEN, replace_token)

    return sources


def preprocess_llama_2(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
    has_image: bool = False
) -> Dict:
    conv = conversation_lib.default_conversation.copy()
    roles = {"human": conv.roles[0], "gpt": conv.roles[1], "model": conv.roles[1]}

    # Apply prompt templates
    conversations = []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != conv.roles[0]:
            # Skip the first one if it is not from human
            source = source[1:]

        conv.messages = []
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            assert role == conv.roles[j % 2], f"{i}"
            conv.append_message(role, sentence["value"])
        conversations.append(conv.get_prompt())

    # Tokenize conversations

    if has_image:
        input_ids = torch.stack([tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations], dim=0)
    else:
        input_ids = tokenizer(
            conversations,
            return_tensors="pt",
            padding="longest",
            max_length=tokenizer.model_max_length,
            truncation=True,
        ).input_ids

    targets = input_ids.clone()

    assert conv.sep_style == conversation_lib.SeparatorStyle.LLAMA_2

    # Mask targets
    sep = "[/INST] "
    for conversation, target in zip(conversations, targets):
        total_len = int(target.ne(tokenizer.pad_token_id).sum())

        rounds = conversation.split(conv.sep2)
        cur_len = 1
        target[:cur_len] = IGNORE_INDEX
        for i, rou in enumerate(rounds):
            if rou == "":
                break

            parts = rou.split(sep)
            if len(parts) != 2:
                break
            parts[0] += sep

            if has_image:
                round_len = len(tokenizer_image_token(rou, tokenizer))
                instruction_len = len(tokenizer_image_token(parts[0], tokenizer)) - 2
            else:
                round_len = len(tokenizer(rou).input_ids)
                instruction_len = len(tokenizer(parts[0]).input_ids) - 2

            target[cur_len : cur_len + instruction_len] = IGNORE_INDEX

            cur_len += round_len
        target[cur_len:] = IGNORE_INDEX

        if cur_len < tokenizer.model_max_length:
            if cur_len != total_len:
                target[:] = IGNORE_INDEX
                print(
                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
                    f" (ignored)"
                )

    return dict(
        input_ids=input_ids,
        labels=targets,
    )


def preprocess_v1(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
    has_image: bool = False
) -> Dict:
    conv = conversation_lib.default_conversation.copy()
    roles = {"human": conv.roles[0], "gpt": conv.roles[1], "model": conv.roles[1]}

    # Apply prompt templates
    conversations = []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != conv.roles[0]:
            # Skip the first one if it is not from human
            source = source[1:]

        conv.messages = []
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            assert role == conv.roles[j % 2], f"{i}"
            conv.append_message(role, sentence["value"])
        conversations.append(conv.get_prompt())

    # Tokenize conversations

    if has_image:
        input_ids = torch.stack([tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations], dim=0)
    else:
        input_ids = tokenizer(
            conversations,
            return_tensors="pt",
            padding="longest",
            max_length=tokenizer.model_max_length,
            truncation=True,
        ).input_ids

    targets = input_ids.clone()

    assert conv.sep_style == conversation_lib.SeparatorStyle.TWO

    # Mask targets
    sep = conv.sep + conv.roles[1] + ": "
    for conversation, target in zip(conversations, targets):
        total_len = int(target.ne(tokenizer.pad_token_id).sum())

        rounds = conversation.split(conv.sep2)
        cur_len = 1
        target[:cur_len] = IGNORE_INDEX
        for i, rou in enumerate(rounds):
            if rou == "":
                break

            parts = rou.split(sep)
            if len(parts) != 2:
                break
            parts[0] += sep

            if has_image:
                round_len = len(tokenizer_image_token(rou, tokenizer))
                instruction_len = len(tokenizer_image_token(parts[0], tokenizer)) - 2
            else:
                round_len = len(tokenizer(rou).input_ids)
                instruction_len = len(tokenizer(parts[0]).input_ids) - 2

            if i != 0 and not tokenizer.legacy and IS_TOKENIZER_GREATER_THAN_0_14:
                round_len -= 1
                instruction_len -= 1

            target[cur_len : cur_len + instruction_len] = IGNORE_INDEX

            cur_len += round_len
        target[cur_len:] = IGNORE_INDEX

        if cur_len < tokenizer.model_max_length:
            if cur_len != total_len:
                target[:] = IGNORE_INDEX
                print(
                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
                    f" (ignored)"
                )

    return dict(
        input_ids=input_ids,
        labels=targets,
    )


def preprocess_mpt(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
    has_image: bool = False
) -> Dict:
    conv = conversation_lib.default_conversation.copy()
    roles = {"human": conv.roles[0], "gpt": conv.roles[1], "model": conv.roles[1]}

    # Apply prompt templates
    conversations = []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != conv.roles[0]:
            # Skip the first one if it is not from human
            source = source[1:]

        conv.messages = []
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            assert role == conv.roles[j % 2], f"{i}"
            conv.append_message(role, sentence["value"])
        conversations.append(conv.get_prompt())

    # Tokenize conversations

    if has_image:
        input_ids = torch.stack([tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations], dim=0)
    else:
        input_ids = tokenizer(
            conversations,
            return_tensors="pt",
            padding="longest",
            max_length=tokenizer.model_max_length,
            truncation=True,
        ).input_ids

    targets = input_ids.clone()
    assert conv.sep_style == conversation_lib.SeparatorStyle.MPT

    # Mask targets
    sep = conv.sep + conv.roles[1]
    for conversation, target in zip(conversations, targets):
        total_len = int(target.ne(tokenizer.pad_token_id).sum())

        rounds = conversation.split(conv.sep)
        re_rounds = [conv.sep.join(rounds[:3])] # system + user + gpt
        for conv_idx in range(3, len(rounds), 2):
            re_rounds.append(conv.sep.join(rounds[conv_idx:conv_idx+2]))    # user + gpt
        cur_len = 0
        target[:cur_len] = IGNORE_INDEX
        for i, rou in enumerate(re_rounds):
            if rou == "":
                break

            parts = rou.split(sep)
            if len(parts) != 2:
                break
            parts[0] += sep

            if has_image:
                round_len = len(tokenizer_image_token(rou, tokenizer))
                instruction_len = len(tokenizer_image_token(parts[0], tokenizer)) - 1
            else:
                round_len = len(tokenizer(rou).input_ids)
                instruction_len = len(tokenizer(parts[0]).input_ids) - 1

            if i != 0 and getattr(tokenizer, 'legacy', False) and IS_TOKENIZER_GREATER_THAN_0_14:
                round_len += 1
                instruction_len += 1

            target[cur_len : cur_len + instruction_len] = IGNORE_INDEX

            cur_len += round_len
        target[cur_len:] = IGNORE_INDEX

        if cur_len < tokenizer.model_max_length:
            if cur_len != total_len:
                target[:] = IGNORE_INDEX
                print(
                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
                    f" (ignored)"
                )

    return dict(
        input_ids=input_ids,
        labels=targets,
    )


def preprocess_plain(
    sources: Sequence[str],
    tokenizer: transformers.PreTrainedTokenizer,
) -> Dict:
    # add end signal and concatenate MM_TOKEN_INDEXtogether
    conversations = []
    for source in sources:
        assert len(source) == 2
        # assert DEFAULT_IMAGE_TOKEN in source[0]['value']
        # source[0]['value'] = DEFAULT_IMAGE_TOKEN
        conversation = source[0]['value'] + source[1]['value'] + conversation_lib.default_conversation.sep
        conversations.append(conversation)
    # tokenize conversations
    input_ids = [tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations]
    targets = copy.deepcopy(input_ids)
    for target, source in zip(targets, sources):
        tokenized_len = len(tokenizer_image_token(source[0]['value'], tokenizer))
        target[:tokenized_len] = IGNORE_INDEX

    return dict(input_ids=input_ids, labels=targets)




def preprocess_gemma(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
    has_image: bool = False
) -> Dict:
    conv = conversation_lib.default_conversation.copy()
    roles = {"human": conv.roles[0], "gpt": conv.roles[1], "model": conv.roles[1]}

    # Apply prompt templates
    conversations = []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != conv.roles[0]:
            # Skip the first one if it is not from human
            source = source[1:]

        conv.messages = []
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            assert role == conv.roles[j % 2], f"{i}"
            conv.append_message(role, sentence["value"])
        conversations.append(conv.get_prompt(use_chat_template=True, tokenizer=tokenizer))

    # Tokenize conversations

    if has_image:
        input_ids = torch.stack([tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations], dim=0)
    else:
        input_ids = tokenizer(
            conversations,
            return_tensors="pt",
            padding="longest",
            max_length=tokenizer.model_max_length,
            truncation=True,
        ).input_ids

    targets = input_ids.clone()

    # Mask targets
    sep = conv.sep + conv.roles[1] + '\n'
    sep2 = conv.sep2 + '\n' + conv.sep + conv.roles[0]
    for conversation, target in zip(conversations, targets):
        total_len = int(target.ne(tokenizer.pad_token_id).sum())

        rounds = conversation.split(sep2)
        cur_len = 1
        target[:cur_len] = IGNORE_INDEX
        for i, rou in enumerate(rounds):
            if rou == "":
                break
            if i != len(rounds) - 1:
                rou += conv.sep2 + '\n'
            if i >= 1 :
                rou = conv.sep + conv.roles[0] + rou

            parts = rou.split(sep)
            if len(parts) != 2:
                break
            parts[0] += sep

            if has_image:
                round_len = len(tokenizer_image_token(rou, tokenizer)) - 1
                instruction_len = len(tokenizer_image_token(parts[0], tokenizer)) - 1
            else:
                raise NotImplementedError


            target[cur_len : cur_len + instruction_len] = IGNORE_INDEX

            cur_len += round_len
        target[cur_len:] = IGNORE_INDEX

        if cur_len < tokenizer.model_max_length:
            if cur_len != total_len:
                target[:] = IGNORE_INDEX
                print(
                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
                    f" (ignored)"
                )

    return dict(
        input_ids=input_ids,
        labels=targets,
    )


def preprocess_mistral(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
    has_image: bool = False
) -> Dict:
    conv = conversation_lib.default_conversation.copy()
    roles = {"human": conv.roles[0], "gpt": conv.roles[1], "model": conv.roles[1]}

    # Apply prompt templates
    conversations = []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != conv.roles[0]:
            # Skip the first one if it is not from human
            source = source[1:]

        conv.messages = []
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            assert role == conv.roles[j % 2], f"{i}"
            conv.append_message(role, sentence["value"])
        conversations.append(conv.get_prompt(use_chat_template=True, tokenizer=tokenizer))

    # Tokenize conversations

    if has_image:
        input_ids = torch.stack([tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations], dim=0)
    else:
        input_ids = tokenizer(
            conversations,
            return_tensors="pt",
            padding="longest",
            max_length=tokenizer.model_max_length,
            truncation=True,
        ).input_ids

    targets = input_ids.clone()

    # Mask targets
    sep = " [/INST]"
    for conversation, target in zip(conversations, targets):
        total_len = int(target.ne(tokenizer.pad_token_id).sum())

        rounds = conversation.split(conv.sep2)
        cur_len = 1
        target[:cur_len] = IGNORE_INDEX
        for i, rou in enumerate(rounds):
            if rou == "":
                break
            parts = rou.split(sep)
            if len(parts) != 2:
                break
            parts[0] += sep

            if has_image:
                round_len = len(tokenizer_image_token(rou, tokenizer))
                instruction_len = len(tokenizer_image_token(parts[0], tokenizer)) - 1
            else:
                round_len = len(tokenizer(rou).input_ids)
                instruction_len = len(tokenizer(parts[0]).input_ids) - 1

            target[cur_len : cur_len + instruction_len] = IGNORE_INDEX

            cur_len += round_len
        target[cur_len:] = IGNORE_INDEX
        if rou[-1] == ' ':   # '</s> ' another space after </s>
            cur_len += 1

        if cur_len < tokenizer.model_max_length:
            if cur_len != total_len:
                target[:] = IGNORE_INDEX
                print(
                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
                    f" (ignored)"
                )

    return dict(
        input_ids=input_ids,
        labels=targets,
    )


def preprocess_thoth(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
    has_image: bool = False
) -> Dict:
    conv = conversation_lib.default_conversation.copy()
    roles = {"human": conv.roles[0], "gpt": conv.roles[1], "model": conv.roles[1]}

    # Apply prompt templates
    conversations = []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != conv.roles[0]:
            # Skip the first one if it is not from human
            source = source[1:]

        conv.messages = []
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            assert role == conv.roles[j % 2], f"{i}"
            conv.append_message(role, sentence["value"])
        conversations.append(conv.get_prompt())

    # Tokenize conversations

    if has_image:
        input_ids = torch.stack([tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations], dim=0)
    else:
        input_ids = tokenizer(
            conversations,
            return_tensors="pt",
            padding="longest",
            max_length=tokenizer.model_max_length,
            truncation=True,
        ).input_ids

    targets = input_ids.clone()


    # Mask targets
    sep = conv.sep + conv.roles[1] + ": "
    for conversation, target in zip(conversations, targets):
        total_len = int(target.ne(tokenizer.pad_token_id).sum())

        rounds = conversation.split(conv.sep2)
        cur_len = 1
        target[:cur_len] = IGNORE_INDEX
        for i, rou in enumerate(rounds):
            if rou == "":
                break

            parts = rou.split(sep)
            if len(parts) != 2:
                break
            parts[0] += sep

            if has_image:
                round_len = len(tokenizer_image_token(rou, tokenizer))
                instruction_len = len(tokenizer_image_token(parts[0], tokenizer)) - 2
            else:
                round_len = len(tokenizer(rou).input_ids)
                instruction_len = len(tokenizer(parts[0]).input_ids) - 2

            target[cur_len: cur_len + instruction_len] = IGNORE_INDEX
            cur_len += round_len + 1
            if i == 0:
                cur_len -= 1
        target[cur_len:] = IGNORE_INDEX

        if cur_len < tokenizer.model_max_length:
            if cur_len != total_len:
                target[:] = IGNORE_INDEX
                print(
                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
                    f" (ignored)"
                )

    return dict(
        input_ids=input_ids,
        labels=targets,
    )


def preprocess(
    sources: Sequence[str],
    tokenizer: transformers.PreTrainedTokenizer,
    has_image: bool = False
) -> Dict:
    """
    Given a list of sources, each is a conversation list. This transform:
    1. Add signal '### ' at the beginning each sentence, with end signal '\n';
    2. Concatenate conversations together;
    3. Tokenize the concatenated conversation;
    4. Make a deepcopy as the target. Mask human words with IGNORE_INDEX.
    """
    if conversation_lib.default_conversation.sep_style == conversation_lib.SeparatorStyle.PLAIN:
        return preprocess_plain(sources, tokenizer)
    if conversation_lib.default_conversation.sep_style == conversation_lib.SeparatorStyle.LLAMA_2:
        return preprocess_llama_2(sources, tokenizer, has_image=has_image)
    if conversation_lib.default_conversation.version.startswith("v1"):
        return preprocess_v1(sources, tokenizer, has_image=has_image)
    if conversation_lib.default_conversation.version == "mpt":
        return preprocess_mpt(sources, tokenizer, has_image=has_image)
    if conversation_lib.default_conversation.version == 'gemma':
        return preprocess_gemma(sources, tokenizer, has_image=has_image)
    if conversation_lib.default_conversation.version == 'thoth':
        return preprocess_thoth(sources, tokenizer, has_image=has_image)
    if conversation_lib.default_conversation.version == 'mistral':
        return preprocess_mistral(sources, tokenizer, has_image=has_image)
    # add end signal and concatenate together
    conversations = []
    for source in sources:
        header = f"{conversation_lib.default_conversation.system}\n\n"
        conversation = _add_speaker_and_signal(header, source)
        conversations.append(conversation)
    # tokenize conversations
    def get_tokenize_len(prompts):
        return [len(tokenizer_image_token(prompt, tokenizer)) for prompt in prompts]

    if has_image:
        input_ids = [tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations]
    else:
        conversations_tokenized = _tokenize_fn(conversations, tokenizer)
        input_ids = conversations_tokenized["input_ids"]

    targets = copy.deepcopy(input_ids)
    for target, source in zip(targets, sources):
        if has_image:
            tokenized_lens = get_tokenize_len([header] + [s["value"] for s in source])
        else:
            tokenized_lens = _tokenize_fn([header] + [s["value"] for s in source], tokenizer)["input_ids_lens"]
        speakers = [sentence["from"] for sentence in source]
        _mask_targets(target, tokenized_lens, speakers)

    return dict(input_ids=input_ids, labels=targets)