[Feat] Support Qwen-VL-Chat on MMBench. (#312)

* [Feat] Support Qwen-VL base.

* [Feat] Support Qwen-VL-Chat on MMBench.

* [Fix] Add postprocessor and fix format.

* [Fix] Add type hint and remove redundant codes.

* [Fix] fix bugs in postprocessor.

* [Fix] Use given commit id.

configs/multimodal/qwen/qwenvl_base_7b_mmbench.py

@@ -0,0 +1,41 @@
+from opencompass.multimodal.models.qwen import QwenVLMMBenchPromptConstructor, QwenVLBasePostProcessor
+
+# dataloader settings
+val_pipeline = [
+    dict(type='mmpretrain.torchvision/Resize',
+         size=(448, 448),
+         interpolation=3),
+    dict(type='mmpretrain.torchvision/ToTensor'),
+    dict(type='mmpretrain.torchvision/Normalize',
+         mean=(0.48145466, 0.4578275, 0.40821073),
+         std=(0.26862954, 0.26130258, 0.27577711)),
+    dict(type='mmpretrain.PackInputs',
+         algorithm_keys=[
+             'question', 'options', 'category', 'l2-category', 'context',
+             'index', 'options_dict'
+         ])
+]
+
+dataset = dict(type='opencompass.MMBenchDataset',
+               data_file='data/mmbench/mmbench_test_20230712.tsv',
+               pipeline=val_pipeline)
+
+qwen_mmbench_dataloader = dict(batch_size=1,
+                  num_workers=4,
+                  dataset=dataset,
+                  collate_fn=dict(type='pseudo_collate'),
+                  sampler=dict(type='DefaultSampler', shuffle=False))
+
+# model settings
+qwen_model = dict(
+    type='qwen-vl-base',
+    pretrained_path='Qwen/Qwen-VL',  # or Huggingface repo id
+    prompt_constructor=dict(type=QwenMMBenchPromptConstructor),
+    post_processor=dict(type=QwenVLBasePostProcessor)
+)
+
+# evaluation settings
+qwen_mmbench_evaluator = [
+    dict(type='opencompass.DumpResults',
+         save_path='work_dirs/qwenvl-base-7b-mmbench.xlsx')
+]

configs/multimodal/qwen/qwenvl_chat_7b_mmbench.py

@@ -0,0 +1,40 @@
+from opencompass.multimodal.models.qwen import QwenVLMMBenchPromptConstructor
+
+# dataloader settings
+val_pipeline = [
+    dict(type='mmpretrain.torchvision/Resize',
+         size=(448, 448),
+         interpolation=3),
+    dict(type='mmpretrain.torchvision/ToTensor'),
+    dict(type='mmpretrain.torchvision/Normalize',
+         mean=(0.48145466, 0.4578275, 0.40821073),
+         std=(0.26862954, 0.26130258, 0.27577711)),
+    dict(type='mmpretrain.PackInputs',
+         algorithm_keys=[
+             'question', 'options', 'category', 'l2-category', 'context',
+             'index', 'options_dict'
+         ])
+]
+
+dataset = dict(type='opencompass.MMBenchDataset',
+               data_file='data/mmbench/mmbench_test_20230712.tsv',
+               pipeline=val_pipeline)
+
+qwen_mmbench_dataloader = dict(batch_size=1,
+                  num_workers=4,
+                  dataset=dataset,
+                  collate_fn=dict(type='pseudo_collate'),
+                  sampler=dict(type='DefaultSampler', shuffle=False))
+
+# model settings
+qwen_model = dict(
+    type='qwen-vl-chat',
+    pretrained_path='Qwen/Qwen-VL-Chat',  # or Huggingface repo id
+    prompt_constructor=dict(type=QwenVLMMBenchPromptConstructor)
+)
+
+# evaluation settings
+qwen_mmbench_evaluator = [
+    dict(type='opencompass.DumpResults',
+         save_path='work_dirs/qwenvl-chat-7b-mmbench.xlsx')
+]

opencompass/multimodal/models/__init__.py

@@ -23,4 +23,5 @@ from .openflamingo import *  # noqa: F401, F403
 if osp.exists('opencompass/multimodal/models/otter/Otter'):
     from .otter import *  # noqa: F401, F403
 
+from .qwen import *  # noqa: F401, F403
 from .visualglm import *  # noqa: F401, F403

opencompass/multimodal/models/qwen/__init__.py

@@ -0,0 +1,8 @@
+from .post_processor import QwenVLBasePostProcessor
+from .prompt_constructor import QwenVLMMBenchPromptConstructor
+from .qwen import QwenVLBase, QwenVLChat
+
+__all__ = [
+    'QwenVLBase', 'QwenVLChat', 'QwenVLBasePostProcessor',
+    'QwenVLMMBenchPromptConstructor'
+]

opencompass/multimodal/models/qwen/generation_utils.py

@@ -0,0 +1,293 @@
+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+"""Generation support."""
+
+from typing import List, Tuple, Union
+
+import torch
+from transformers import PreTrainedTokenizer
+
+# Types.
+HistoryType = List[Tuple[str, str]]
+TokensType = List[int]
+BatchTokensType = List[List[int]]
+
+
+def pad_batch(batch: BatchTokensType, pad_id: int,
+              seq_length: int) -> BatchTokensType:
+    for tokens in batch:
+        context_length = len(tokens)
+        if context_length < seq_length:
+            tokens.extend([pad_id] * (seq_length - context_length))
+    return batch
+
+
+def get_ltor_masks_and_position_ids(
+    data: torch.Tensor,
+    eod_token: int,
+    reset_position_ids: bool,
+    reset_attention_mask: bool,
+    eod_mask_loss: bool,
+):
+    """Build masks and position id for left to right model."""
+
+    # Extract batch size and sequence length.
+    micro_batch_size, seq_length = data.size()
+
+    # Attention mask (lower triangular).
+    if reset_attention_mask:
+        att_mask_batch = micro_batch_size
+    else:
+        att_mask_batch = 1
+    attention_mask = torch.tril(
+        torch.ones((att_mask_batch, seq_length, seq_length),
+                   device=data.device)).view(att_mask_batch, 1, seq_length,
+                                             seq_length)
+
+    # Loss mask.
+    loss_mask = torch.ones(data.size(), dtype=torch.float, device=data.device)
+    if eod_mask_loss:
+        loss_mask[data == eod_token] = 0.0
+
+    # Position ids.
+    position_ids = torch.arange(seq_length,
+                                dtype=torch.long,
+                                device=data.device)
+    position_ids = position_ids.unsqueeze(0).expand_as(data)
+    # We need to clone as the ids will be modified based on batch index.
+    if reset_position_ids:
+        position_ids = position_ids.clone()
+
+    if reset_position_ids or reset_attention_mask:
+        # Loop through the batches:
+        for b in range(micro_batch_size):
+
+            # Find indices where EOD token is.
+            eod_index = position_ids[b, data[b] == eod_token]
+            # Detach indices from positions if going to modify positions.
+            if reset_position_ids:
+                eod_index = eod_index.clone()
+
+            # Loop through EOD indices:
+            prev_index = 0
+            for j in range(eod_index.size()[0]):
+                i = eod_index[j]
+                # Mask attention loss.
+                if reset_attention_mask:
+                    attention_mask[b, 0, (i + 1):, :(i + 1)] = 0
+                # Reset positions.
+                if reset_position_ids:
+                    position_ids[b, (i + 1):] -= i + 1 - prev_index
+                    prev_index = i + 1
+
+    # Convert attention mask to binary:
+    attention_mask = attention_mask < 0.5
+
+    return attention_mask, loss_mask, position_ids
+
+
+def get_batch(context_tokens: torch.LongTensor, eod_id: int):
+    """Generate batch from context tokens."""
+    # Move to GPU.
+    tokens = context_tokens.contiguous().to(context_tokens.device)
+    # Get the attention mask and position ids.
+    attention_mask, _, position_ids = get_ltor_masks_and_position_ids(
+        tokens,
+        eod_id,
+        reset_position_ids=False,
+        reset_attention_mask=False,
+        eod_mask_loss=False,
+    )
+    return tokens, attention_mask, position_ids
+
+
+def get_stop_words_ids(chat_format: str, tokenizer: PreTrainedTokenizer):
+    if chat_format == 'raw':
+        stop_words_ids = [tokenizer.encode('Human:'), [tokenizer.eod_id]]
+    elif chat_format == 'chatml':
+        stop_words_ids = [[tokenizer.im_end_id], [tokenizer.im_start_id]]
+    else:
+        raise NotImplementedError(f'Unknown chat format {chat_format!r}')
+    return stop_words_ids
+
+
+def make_context(
+    tokenizer: PreTrainedTokenizer,
+    query: str,
+    history: List[Tuple[str, str]] = None,
+    system: str = '',
+    max_window_size: int = 6144,
+    chat_format: str = 'chatml',
+):
+    if history is None:
+        history = []
+
+    if chat_format == 'chatml':
+        im_start, im_end = '<|im_start|>', '<|im_end|>'
+        im_start_tokens = [tokenizer.im_start_id]
+        im_end_tokens = [tokenizer.im_end_id]
+        nl_tokens = tokenizer.encode('\n')
+
+        def _tokenize_str(role, content):
+            return f'{role}\n{content}', tokenizer.encode(
+                role, allowed_special=set(
+                    tokenizer.IMAGE_ST)) + nl_tokens + tokenizer.encode(
+                        content, allowed_special=set(tokenizer.IMAGE_ST))
+
+        system_text, system_tokens_part = _tokenize_str('system', system)
+        system_tokens = im_start_tokens + system_tokens_part + im_end_tokens
+
+        raw_text = ''
+        context_tokens = []
+
+        for turn_query, turn_response in reversed(history):
+            query_text, query_tokens_part = _tokenize_str('user', turn_query)
+            query_tokens = im_start_tokens + query_tokens_part + im_end_tokens
+            if turn_response is not None:
+                response_text, response_tokens_part = _tokenize_str(
+                    'assistant', turn_response)
+                response_tokens = im_start_tokens + response_tokens_part + im_end_tokens  # noqa
+
+                next_context_tokens = nl_tokens + query_tokens + nl_tokens + response_tokens  # noqa
+                prev_chat = (
+                    f'\n{im_start}{query_text}{im_end}\n{im_start}{response_text}{im_end}'  # noqa
+                )
+            else:
+                next_context_tokens = nl_tokens + query_tokens + nl_tokens
+                prev_chat = f'\n{im_start}{query_text}{im_end}\n'
+
+            current_context_size = (len(system_tokens) +
+                                    len(next_context_tokens) +
+                                    len(context_tokens))
+            if current_context_size < max_window_size:
+                context_tokens = next_context_tokens + context_tokens
+                raw_text = prev_chat + raw_text
+            else:
+                break
+
+        context_tokens = system_tokens + context_tokens
+        raw_text = f'{im_start}{system_text}{im_end}' + raw_text
+        context_tokens += (nl_tokens + im_start_tokens +
+                           _tokenize_str('user', query)[1] + im_end_tokens +
+                           nl_tokens + im_start_tokens +
+                           tokenizer.encode('assistant') + nl_tokens)
+        raw_text += f'\n{im_start}user\n{query}{im_end}\n{im_start}assistant\n'
+
+    elif chat_format == 'raw':
+        raw_text = query
+        context_tokens = tokenizer.encode(raw_text)
+    else:
+        raise NotImplementedError(f'Unknown chat format {chat_format!r}')
+
+    return raw_text, context_tokens
+
+
+def _decode_default(
+    tokens: List[int],
+    *,
+    stop_words: List[str],
+    eod_words: List[str],
+    tokenizer: PreTrainedTokenizer,
+    raw_text_len: int,
+    verbose: bool = False,
+    return_end_reason: bool = False,
+    errors: str = 'replace',
+):
+    trim_decode_tokens = tokenizer.decode(tokens, errors=errors)[raw_text_len:]
+    if verbose:
+        print('\nRaw Generate: ', trim_decode_tokens)
+
+    end_reason = f'Gen length {len(tokens)}'
+    for stop_word in stop_words:
+        trim_decode_tokens = trim_decode_tokens.replace(stop_word, '').strip()
+    for eod_word in eod_words:
+        if eod_word in trim_decode_tokens:
+            end_reason = f'Gen {eod_word!r}'
+        trim_decode_tokens = trim_decode_tokens.split(eod_word)[0]
+    trim_decode_tokens = trim_decode_tokens.strip()
+    if verbose:
+        print('\nEnd Reason:', end_reason)
+        print('\nGenerate: ', trim_decode_tokens)
+
+    if return_end_reason:
+        return trim_decode_tokens, end_reason
+    else:
+        return trim_decode_tokens
+
+
+def _decode_chatml(tokens: List[int],
+                   *,
+                   stop_words: List[str],
+                   eod_token_ids: List[int],
+                   tokenizer: PreTrainedTokenizer,
+                   raw_text_len: int,
+                   context_length: int,
+                   verbose: bool = False,
+                   return_end_reason: bool = False,
+                   errors: str = 'replace'):
+    end_reason = f'Gen length {len(tokens)}'
+    eod_token_idx = context_length
+    for eod_token_idx in range(context_length, len(tokens)):
+        if tokens[eod_token_idx] in eod_token_ids:
+            end_reason = f'Gen {tokenizer.decode([tokens[eod_token_idx]])!r}'
+            break
+
+    trim_decode_tokens = tokenizer.decode(tokens[:eod_token_idx],
+                                          errors=errors)[raw_text_len:]
+    if verbose:
+        print('\nRaw Generate w/o EOD:',
+              tokenizer.decode(tokens, errors=errors)[raw_text_len:])
+        print('\nRaw Generate:', trim_decode_tokens)
+        print('\nEnd Reason:', end_reason)
+    for stop_word in stop_words:
+        trim_decode_tokens = trim_decode_tokens.replace(stop_word, '').strip()
+    trim_decode_tokens = trim_decode_tokens.strip()
+    if verbose:
+        print('\nGenerate:', trim_decode_tokens)
+
+    if return_end_reason:
+        return trim_decode_tokens, end_reason
+    else:
+        return trim_decode_tokens
+
+
+def decode_tokens(
+    tokens: Union[torch.LongTensor, TokensType],
+    tokenizer: PreTrainedTokenizer,
+    raw_text_len: int,
+    context_length: int,
+    chat_format: str,
+    verbose: bool = False,
+    return_end_reason: bool = False,
+    errors: str = 'replace',
+) -> str:
+    if torch.is_tensor(tokens):
+        tokens = tokens.cpu().numpy().tolist()
+
+    if chat_format == 'chatml':
+        return _decode_chatml(
+            tokens,
+            stop_words=[],
+            eod_token_ids=[tokenizer.im_start_id, tokenizer.im_end_id],
+            tokenizer=tokenizer,
+            raw_text_len=raw_text_len,
+            context_length=context_length,
+            verbose=verbose,
+            return_end_reason=return_end_reason,
+            errors=errors,
+        )
+    elif chat_format == 'raw':
+        return _decode_default(
+            tokens,
+            stop_words=['<|endoftext|>'],
+            eod_words=['<|endoftext|>'],
+            tokenizer=tokenizer,
+            raw_text_len=raw_text_len,
+            verbose=verbose,
+            return_end_reason=return_end_reason,
+            errors=errors,
+        )
+    else:
+        raise NotImplementedError(f'Unknown chat format {chat_format!r}')

opencompass/multimodal/models/qwen/post_processor.py

@@ -0,0 +1,16 @@
+from typing import Any
+
+import torch
+
+
+class QwenVLBasePostProcessor:
+    """Post processor for Qwen-VL-Base."""
+
+    def __init__(self) -> None:
+        pass
+
+    def __call__(self, pred: torch.tensor, tokenizer: Any,
+                 input_len: int) -> str:
+        response = self.tokenizer.decode(pred)[input_len:]
+        response = response.replace('<|endoftext|>', '').strip()
+        return response

opencompass/multimodal/models/qwen/prompt_constructor.py

@@ -0,0 +1,29 @@
+class QwenVLMMBenchPromptConstructor:
+    """MMBench prompt constructor for Qwen-VL.
+
+    The output is a dict following the input format of Qwen-VL tokenizer.
+    """
+
+    def __init__(self) -> None:
+        pass
+
+    def __call__(self, inputs: dict) -> str:
+        data_samples = inputs['data_samples']
+        assert len(data_samples) == 1
+        data_sample = data_samples[0]
+        question = data_sample.get('question')
+        options = data_sample.get('options')
+        context = data_sample.get('context')
+        if context is not None:
+            prompt = context + ' ' + question + ' ' + options
+        else:
+            prompt = question + ' ' + options
+        format_input = [
+            {
+                'image': 'This_is_path_to_an_image.'
+            },  # Just placeholder for Image Tokens
+            {
+                'text': prompt
+            },
+        ]
+        return format_input

opencompass/multimodal/models/qwen/qwen.py

@@ -0,0 +1,324 @@
+import types
+from typing import Optional, Tuple
+
+import mmengine
+import torch
+import torch.nn as nn
+from mmengine.device import get_device
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from transformers.generation import GenerationConfig
+from transformers.modeling_outputs import BaseModelOutputWithPast
+
+from opencompass.registry import MM_MODELS
+
+from .generation_utils import decode_tokens, make_context
+
+
+@MM_MODELS.register_module('qwen-vl-base')
+class QwenVLBase(nn.Module):
+    """Inference code of Qwen-VL.
+
+    We load the Qwen model via Huggingface.
+    Args:
+        pretrained_path (str): Path to Qwen checkpoint or repo id.
+        prompt_constructor (dict): The config of prompt constructor.
+        post_processor (dict): The config of post processor.
+        is_caption_task (bool): Whether the task is caption task.
+            Defaults to False.
+        commit_id (str): Use given version of Qwen-VL.
+            Warning: the latest version may have some conflicts.
+            Recommend to use the given default version.
+    """
+
+    def __init__(
+            self,
+            pretrained_path: str,
+            prompt_constructor: dict = None,
+            post_processor: dict = None,
+            is_caption_task: bool = False,
+            commit_id: str = '548275c8b99de56dec203c0e793be18e030f2f4c'
+    ) -> None:
+        super().__init__()
+        self.tokenizer = AutoTokenizer.from_pretrained(pretrained_path,
+                                                       trust_remote_code=True,
+                                                       revision=commit_id)
+        self.model = AutoModelForCausalLM.from_pretrained(
+            pretrained_path,
+            device_map=get_device(),
+            trust_remote_code=True,
+            revision=commit_id)
+        self.model.generation_config = GenerationConfig.from_pretrained(
+            pretrained_path, trust_remote_code=True, revision=commit_id)
+        if prompt_constructor is not None:
+            self.prompt_constructor = mmengine.registry.build_from_cfg(
+                prompt_constructor, MM_MODELS)
+        if post_processor is not None:
+            self.post_processor = mmengine.registry.build_from_cfg(
+                post_processor, MM_MODELS)
+        self.is_caption_task = is_caption_task
+        self.model.transformer.forward = types.MethodType(
+            forward_hack, self.model.transformer)
+
+    def _build_embeds(self, images, input_ids):
+        # encode image
+        images = self.model.transformer.visual(images)
+        # compute image position
+        bos_pos = torch.where(input_ids == self.model.transformer.config.
+                              visual['image_start_id'])
+        eos_pos = torch.where(
+            input_ids ==
+            self.model.transformer.config.visual['image_start_id'] + 1)
+        assert (bos_pos[0] == eos_pos[0]).all()
+        img_pos = torch.stack((bos_pos[0], bos_pos[1], eos_pos[1]), dim=1)
+        # embed words
+        inputs_embeds = self.model.transformer.wte(input_ids)
+        # embed image tokens
+        for idx, (i, a, b) in enumerate(img_pos):
+            inputs_embeds[i][a + 1:b] = images[idx]
+        return inputs_embeds
+
+    def generate(self, batch):
+        images = batch.pop('inputs')
+        images = torch.stack(images, dim=0)
+        format_input = self.prompt_constructor(batch)
+        query = self.tokenizer.from_list_format(format_input)
+
+        inputs = self.tokenizer(query, return_tensors='pt')
+        inputs = inputs.to(get_device())
+        input_ids, token_type_ids, attention_mask = inputs[
+            'input_ids'], inputs['token_type_ids'], inputs['attention_mask']
+        inputs_embeds = self._build_embeds(images, input_ids)
+        pred = self.model.generate(input_ids=input_ids,
+                                   inputs_embeds=inputs_embeds,
+                                   attention_mask=attention_mask,
+                                   token_type_ids=token_type_ids)
+        response = self.post_processor(pred.cpu()[0])
+
+        data_sample = batch['data_samples'][0]
+        if self.is_caption_task:
+            data_sample.pred_caption = response
+        else:
+            data_sample.pred_answer = response
+        return data_sample
+
+    def forward(self, batch):
+        return self.generate(batch)
+
+
+@MM_MODELS.register_module('qwen-vl-chat')
+class QwenVLChat(QwenVLBase):
+    """Inference code of Qwen-VL-Chat.
+
+    We load the Qwen model via Huggingface.
+    Args:
+        pretrained_path (str): Path to Qwen checkpoint or repo id.
+        prompt_constructor (dict): The config of prompt constructor.
+        post_processor (dict): The config of post processor.
+        is_caption_task (bool): Whether the task is caption task.
+            Defaults to False.
+    """
+
+    def __init__(self,
+                 pretrained_path: str,
+                 prompt_constructor: dict = None,
+                 post_processor: dict = None,
+                 is_caption_task: bool = False) -> None:
+        super().__init__(pretrained_path, prompt_constructor, post_processor,
+                         is_caption_task)
+
+    def generate(self, batch):
+        images = batch.pop('inputs')
+        images = torch.stack(images, dim=0)
+        format_input = self.prompt_constructor(batch)
+        query = self.tokenizer.from_list_format(format_input)
+
+        raw_text, context_tokens = make_context(
+            self.tokenizer,
+            query,
+            system='You are a helpful assistant.',
+            chat_format=self.model.generation_config.chat_format,
+        )
+
+        input_ids = torch.tensor([context_tokens]).to(get_device())
+
+        inputs_embeds = self._build_embeds(images, input_ids)
+        pred = self.model.generate(input_ids=input_ids,
+                                   inputs_embeds=inputs_embeds)
+
+        response = decode_tokens(
+            pred[0],
+            self.tokenizer,
+            raw_text_len=len(raw_text),
+            context_length=len(context_tokens),
+            chat_format=self.model.generation_config.chat_format,
+            verbose=False,
+            errors='replace')
+
+        data_sample = batch['data_samples'][0]
+        if self.is_caption_task:
+            data_sample.pred_caption = response
+        else:
+            data_sample.pred_answer = response
+        return data_sample
+
+
+def forward_hack(self,
+                 input_ids: Optional[torch.LongTensor] = None,
+                 past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
+                 attention_mask: Optional[torch.FloatTensor] = None,
+                 token_type_ids: Optional[torch.LongTensor] = None,
+                 position_ids: Optional[torch.LongTensor] = None,
+                 head_mask: Optional[torch.FloatTensor] = None,
+                 inputs_embeds: Optional[torch.FloatTensor] = None,
+                 encoder_hidden_states: Optional[torch.Tensor] = None,
+                 encoder_attention_mask: Optional[torch.FloatTensor] = None,
+                 use_cache: Optional[bool] = None,
+                 output_attentions: Optional[bool] = None,
+                 output_hidden_states: Optional[bool] = None,
+                 return_dict: Optional[bool] = None):
+    if past_key_values is None and input_ids is not None and torch.any(
+            input_ids == self.config.visual['image_start_id']):
+        bos_pos = torch.where(
+            input_ids == self.config.visual['image_start_id'])
+        eos_pos = torch.where(
+            input_ids == self.config.visual['image_start_id'] + 1)
+        assert (bos_pos[0] == eos_pos[0]).all()
+        img_pos = torch.stack((bos_pos[0], bos_pos[1], eos_pos[1]), dim=1)
+        images = []
+        for i, a, b in img_pos:
+            image = input_ids[i][a + 1:b - 1].tolist()
+            image = image[:image.index(self.config.visual['image_start_id'] +
+                                       2)]
+            images.append(bytes(image).decode('utf-8'))
+
+        images = self.visual.encode(images)
+        assert images.shape[0] == len(images)
+    else:
+        images = None
+
+    output_attentions = (output_attentions if output_attentions is not None
+                         else self.config.output_attentions)
+    output_hidden_states = (output_hidden_states if output_hidden_states
+                            is not None else self.config.output_hidden_states)
+    use_cache = use_cache if use_cache is not None else self.config.use_cache
+    return_dict = (return_dict
+                   if return_dict is not None else self.config.use_return_dict)
+
+    if input_ids is not None and inputs_embeds is not None:
+        raise ValueError(
+            'You cannot specify both input_ids and inputs_embeds at the same time'  # noqa
+        )
+    elif input_ids is not None:
+        input_shape = input_ids.size()
+        input_ids = input_ids.view(-1, input_shape[-1])
+        batch_size = input_ids.shape[0]
+    elif inputs_embeds is not None:
+        input_shape = inputs_embeds.size()[:-1]
+        batch_size = inputs_embeds.shape[0]
+    else:
+        raise ValueError(
+            'You have to specify either input_ids or inputs_embeds')
+
+    device = input_ids.device if input_ids is not None else inputs_embeds.device  # noqa
+
+    if token_type_ids is not None:
+        token_type_ids = token_type_ids.view(-1, input_shape[-1])
+    if position_ids is not None:
+        position_ids = position_ids.view(-1, input_shape[-1])
+
+    if past_key_values is None:
+        past_length = 0
+        past_key_values = tuple([None] * len(self.h))
+    else:
+        past_length = past_key_values[0][0].size(-2)
+
+    if position_ids is None:
+        position_ids = torch.arange(
+            past_length,
+            input_shape[-1] + past_length,
+            dtype=torch.long,
+            device=device,
+        )
+        position_ids = position_ids.unsqueeze(0).view(-1, input_shape[-1])
+
+    encoder_attention_mask = None
+    head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)
+
+    if inputs_embeds is None:
+        inputs_embeds = self.wte(input_ids)
+
+    if batch_size <= 0:
+        raise ValueError('batch_size has to be defined and > 0')
+    attention_mask = self._prepare_decoder_attention_mask(
+        attention_mask, input_shape, inputs_embeds, past_length)
+
+    hidden_states = inputs_embeds
+
+    hidden_states = self.drop(hidden_states)
+    if images is not None:
+        for idx, (i, a, b) in enumerate(img_pos):
+            hidden_states[i][a + 1:b] = images[idx]
+    output_shape = input_shape + (hidden_states.size(-1), )
+
+    presents = () if use_cache else None
+    all_self_attentions = () if output_attentions else None
+    all_hidden_states = () if output_hidden_states else None
+    for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
+
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states, )
+
+        if self.gradient_checkpointing and self.training:
+
+            def create_custom_forward(module):
+
+                def custom_forward(*inputs):
+                    # None for past_key_value
+                    return module(*inputs, use_cache, output_attentions)
+
+                return custom_forward
+
+            outputs = torch.utils.checkpoint.checkpoint(
+                create_custom_forward(block),
+                hidden_states,
+                None,
+                attention_mask,
+                head_mask[i],
+                encoder_hidden_states,
+                encoder_attention_mask,
+            )
+        else:
+            outputs = block(
+                hidden_states,
+                layer_past=layer_past,
+                attention_mask=attention_mask,
+                head_mask=head_mask[i],
+                encoder_hidden_states=encoder_hidden_states,
+                encoder_attention_mask=encoder_attention_mask,
+                use_cache=use_cache,
+                output_attentions=output_attentions,
+            )
+
+        hidden_states = outputs[0]
+        if use_cache is True:
+            presents = presents + (outputs[2 if output_attentions else 1], )
+
+        if output_attentions:
+            all_self_attentions = all_self_attentions + (outputs[1], )
+
+    hidden_states = self.ln_f(hidden_states)
+    hidden_states = hidden_states.view(output_shape)
+    # Add last hidden state
+    if output_hidden_states:
+        all_hidden_states = all_hidden_states + (hidden_states, )
+
+    if not return_dict:
+        return tuple(v for v in [hidden_states, presents, all_hidden_states]
+                     if v is not None)
+
+    return BaseModelOutputWithPast(
+        last_hidden_state=hidden_states,
+        past_key_values=presents,
+        hidden_states=all_hidden_states,
+        attentions=all_self_attentions,
+    )