Upload 5 files

configuration_mplug_owl2.py

@@ -0,0 +1,334 @@
+# Copyright (c) Alibaba.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+import copy
+import os
+from typing import Union
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.models.auto.modeling_auto import MODEL_FOR_CAUSAL_LM_MAPPING_NAMES
+from transformers.utils import logging
+from transformers.models.auto import CONFIG_MAPPING
+
+
+class LlamaConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`LlamaModel`]. It is used to instantiate an LLaMA
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the LLaMA-7B.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 32000):
+            Vocabulary size of the LLaMA model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`LlamaModel`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 11008):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer decoder.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        num_key_value_heads (`int`, *optional*):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
+            `num_attention_heads`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 2048):
+            The maximum sequence length that this model might ever be used with. Llama 1 supports up to 2048 tokens,
+            Llama 2 up to 4096, CodeLlama up to 16384.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        pad_token_id (`int`, *optional*):
+            Padding token id.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            Beginning of stream token id.
+        eos_token_id (`int`, *optional*, defaults to 2):
+            End of stream token id.
+        pretraining_tp (`int`, *optional*, defaults to 1):
+            Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this
+            document](https://huggingface.co/docs/transformers/parallelism) to understand more about it. This value is
+            necessary to ensure exact reproducibility of the pretraining results. Please refer to [this
+            issue](https://github.com/pytorch/pytorch/issues/76232).
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        rope_scaling (`Dict`, *optional*):
+            Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports two scaling
+            strategies: linear and dynamic. Their scaling factor must be a float greater than 1. The expected format is
+            `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
+            `max_position_embeddings` to the expected new maximum. See the following thread for more information on how
+            these scaling strategies behave:
+            https://www.reddit.com/r/LocalLLaMA/comments/14mrgpr/dynamically_scaled_rope_further_increases/. This is an
+            experimental feature, subject to breaking API changes in future versions.
+        attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`):
+            Whether to use a bias in the query, key, value and output projection layers during self-attention.
+
+
+    ```python
+    >>> from transformers import LlamaModel, LlamaConfig
+
+    >>> # Initializing a LLaMA llama-7b style configuration
+    >>> configuration = LlamaConfig()
+
+    >>> # Initializing a model from the llama-7b style configuration
+    >>> model = LlamaModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+    model_type = "llama"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=32000,
+        hidden_size=4096,
+        intermediate_size=11008,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=None,
+        hidden_act="silu",
+        max_position_embeddings=2048,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=1,
+        eos_token_id=2,
+        pretraining_tp=1,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        rope_scaling=None,
+        attention_bias=False,
+        attention_dropout=0.0,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+
+        # for backward compatibility
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+
+        self.num_key_value_heads = num_key_value_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.pretraining_tp = pretraining_tp
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.rope_scaling = rope_scaling
+        self._rope_scaling_validation()
+        self.attention_bias = attention_bias
+        self.attention_dropout = attention_dropout
+
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+
+    def _rope_scaling_validation(self):
+        """
+        Validate the `rope_scaling` configuration.
+        """
+        if self.rope_scaling is None:
+            return
+
+        if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) != 2:
+            raise ValueError(
+                "`rope_scaling` must be a dictionary with with two fields, `type` and `factor`, "
+                f"got {self.rope_scaling}"
+            )
+        rope_scaling_type = self.rope_scaling.get("type", None)
+        rope_scaling_factor = self.rope_scaling.get("factor", None)
+        if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic"]:
+            raise ValueError(
+                f"`rope_scaling`'s type field must be one of ['linear', 'dynamic'], got {rope_scaling_type}"
+            )
+        if rope_scaling_factor is None or not isinstance(rope_scaling_factor, float) or rope_scaling_factor <= 1.0:
+            raise ValueError(f"`rope_scaling`'s factor field must be a float > 1, got {rope_scaling_factor}")
+
+            
+class MplugOwlVisionConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`MplugOwlVisionModel`]. It is used to instantiate
+    a
+     mPLUG-Owl vision encoder according to the specified arguments, defining the model architecture. Instantiating a
+     configuration defaults will yield a similar configuration to that of the mPLUG-Owl
+     [x-plug/x_plug-llama-7b](https://huggingface.co/x-plug/x_plug-llama-7b) architecture.
+
+     Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+     documentation from [`PretrainedConfig`] for more information.
+
+     Args:
+         hidden_size (`int`, *optional*, defaults to 768):
+             Dimensionality of the encoder layers and the pooler layer.
+         intermediate_size (`int`, *optional*, defaults to 3072):
+             Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
+         num_hidden_layers (`int`, *optional*, defaults to 12):
+             Number of hidden layers in the Transformer encoder.
+         num_attention_heads (`int`, *optional*, defaults to 12):
+             Number of attention heads for each attention layer in the Transformer encoder.
+         image_size (`int`, *optional*, defaults to 224):
+             The size (resolution) of each image.
+         patch_size (`int`, *optional*, defaults to 32):
+             The size (resolution) of each patch.
+         hidden_act (`str` or `function`, *optional*, defaults to `"quick_gelu"`):
+             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+             `"relu"`, `"selu"` and `"gelu_new"` ``"quick_gelu"` are supported.
+         layer_norm_eps (`float`, *optional*, defaults to 1e-5):
+             The epsilon used by the layer normalization layers.
+         attention_dropout (`float`, *optional*, defaults to 0.0):
+             The dropout ratio for the attention probabilities.
+         initializer_range (`float`, *optional*, defaults to 0.02):
+             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+         initializer_factor (`float`, *optional*, defaults to 1):
+             A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
+             testing).
+
+
+     ```"""
+
+    model_type = "mplug_owl_vision_model"
+
+    def __init__(
+        self,
+        hidden_size=1024,
+        intermediate_size=4096,
+        projection_dim=768,
+        num_hidden_layers=24,
+        num_attention_heads=16,
+        num_channels=3,
+        image_size=448,
+        patch_size=14,
+        hidden_act="quick_gelu",
+        layer_norm_eps=1e-6,
+        attention_dropout=0.0,
+        initializer_range=0.02,
+        initializer_factor=1.0,
+        use_flash_attn=False,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.projection_dim = projection_dim
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.num_channels = num_channels
+        self.patch_size = patch_size
+        self.image_size = image_size
+        self.initializer_range = initializer_range
+        self.initializer_factor = initializer_factor
+        self.attention_dropout = attention_dropout
+        self.layer_norm_eps = layer_norm_eps
+        self.hidden_act = hidden_act
+        self.use_flash_attn = use_flash_attn
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the vision config dict if we are loading from MplugOwlConfig
+        if config_dict.get("model_type") == "mplug-owl":
+            config_dict = config_dict["vision_config"]
+
+        if "model_type" in config_dict and hasattr(cls, "model_type") and config_dict["model_type"] != cls.model_type:
+            logger.warning(
+                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
+                f"{cls.model_type}. This is not supported for all configurations of models and can yield errors."
+            )
+
+        return cls.from_dict(config_dict, **kwargs)
+
+
+class MplugOwlVisualAbstractorConfig(PretrainedConfig):
+    model_type = "mplug_owl_visual_abstract"
+
+    def __init__(
+        self,
+        num_learnable_queries=64,
+        hidden_size=1024,
+        num_hidden_layers=6,
+        num_attention_heads=16,
+        intermediate_size=2816,
+        attention_probs_dropout_prob=0.,
+        initializer_range=0.02,
+        layer_norm_eps=1e-6,
+        encoder_hidden_size=1024,
+        grid_size=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.hidden_size = hidden_size
+        self.num_learnable_queries = num_learnable_queries
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.intermediate_size = intermediate_size
+        self.attention_probs_dropout_prob = attention_probs_dropout_prob
+        self.initializer_range = initializer_range
+        self.layer_norm_eps = layer_norm_eps
+        self.encoder_hidden_size = encoder_hidden_size
+        self.grid_size = grid_size if grid_size else 32
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the visual_abstractor config dict if we are loading from MplugOwlConfig
+        if config_dict.get("model_type") == "mplug-owl":
+            config_dict = config_dict["abstractor_config"]
+
+        if "model_type" in config_dict and hasattr(cls, "model_type") and config_dict["model_type"] != cls.model_type:
+            logger.warning(
+                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
+                f"{cls.model_type}. This is not supported for all configurations of models and can yield errors."
+            )
+
+        return cls.from_dict(config_dict, **kwargs)
+
+
+
+DEFAULT_VISUAL_CONFIG = {
+    "visual_model": MplugOwlVisionConfig().to_dict(),
+    "visual_abstractor": MplugOwlVisualAbstractorConfig().to_dict()
+}
+
+class MPLUGOwl2Config(LlamaConfig):
+    model_type = "mplug_owl2"
+    def __init__(self, visual_config=None, **kwargs):
+        if visual_config is None:
+            self.visual_config = DEFAULT_VISUAL_CONFIG
+        else:
+            self.visual_config = visual_config
+        
+        super().__init__(
+            **kwargs,
+        )
+        
+if __name__ == "__main__":
+    print(MplugOwlVisionConfig().to_dict())

modeling_attn_mask_utils.py

@@ -0,0 +1,247 @@
+# Copyright 2023 The HuggingFace 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.
+from typing import List, Optional, Tuple, Union
+
+import torch
+
+
+class AttentionMaskConverter:
+    """
+    A utility attention mask class that allows one to:
+        - Create a causal 4d mask
+        - Create a causal 4d mask with slided window
+        - Convert a 2d attention mask (batch_size, query_length) to a 4d attention mask (batch_size, 1, query_length,
+          key_value_length) that can be multiplied with attention scores
+
+    Parameters:
+        is_causal (`bool`):
+            Whether the attention mask should be a uni-directional (causal) or bi-directional mask.
+
+        sliding_window (`int`, *optional*):
+            Optionally, the sliding window masks can be created if `sliding_window` is defined to a positive integer.
+    """
+
+    def __init__(self, is_causal: bool, sliding_window: Optional[int] = None):
+        self.is_causal = is_causal
+        self.sliding_window = sliding_window
+
+        if self.sliding_window is not None and self.sliding_window <= 0:
+            raise ValueError(
+                f"Make sure that when passing `sliding_window` that its value is a strictly positive integer, not `{self.sliding_window}`"
+            )
+
+    def to_causal_4d(
+        self,
+        batch_size: int,
+        query_length: int,
+        key_value_length: int,
+        dtype: torch.dtype = torch.float32,
+        device: Union[torch.device, "str"] = "cpu",
+    ) -> torch.Tensor:
+        """
+        Creates a causal 4D mask of (bsz, head_dim=1, query_length, key_value_length) shape and adds large negative
+        bias to upper right hand triangular matrix (causal mask).
+        """
+        if not self.is_causal:
+            raise ValueError(f"Please use `to_causal_4d` only if {self.__class__} has `is_causal` set to True.")
+
+        # If shape is not cached, create a new causal mask and cache it
+        input_shape = (batch_size, query_length)
+        past_key_values_length = key_value_length - query_length
+
+        # create causal mask
+        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
+        causal_4d_mask = None
+        if input_shape[-1] > 1 or self.sliding_window is not None:
+            causal_4d_mask = self._make_causal_mask(
+                input_shape,
+                dtype,
+                device=device,
+                past_key_values_length=past_key_values_length,
+                sliding_window=self.sliding_window,
+            )
+
+        return causal_4d_mask
+
+    def to_4d(
+        self,
+        attention_mask_2d: torch.Tensor,
+        query_length: int,
+        key_value_length: Optional[int] = None,
+        dtype: torch.dtype = torch.float32,
+    ) -> torch.Tensor:
+        """
+        Converts 2D attention mask to 4D attention mask by expanding mask to (bsz, head_dim=1, query_length,
+        key_value_length) shape and by adding a large negative bias to not-attended positions. If attention_mask is
+        causal, a causal mask will be added.
+        """
+        input_shape = (attention_mask_2d.shape[0], query_length)
+
+        # create causal mask
+        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
+        causal_4d_mask = None
+        if (input_shape[-1] > 1 or self.sliding_window is not None) and self.is_causal:
+            if key_value_length is None:
+                raise ValueError(
+                    "This attention mask converter is causal. Make sure to pass `key_value_length` to correctly create a causal mask."
+                )
+
+            past_key_values_length = key_value_length - query_length
+            causal_4d_mask = self._make_causal_mask(
+                input_shape,
+                dtype,
+                device=attention_mask_2d.device,
+                past_key_values_length=past_key_values_length,
+                sliding_window=self.sliding_window,
+            )
+        elif self.sliding_window is not None:
+            raise NotImplementedError("Sliding window is currently only implemented for causal masking")
+
+        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
+        expanded_attn_mask = self._expand_mask(attention_mask_2d, dtype, tgt_len=input_shape[-1]).to(
+            attention_mask_2d.device
+        )
+        expanded_4d_mask = expanded_attn_mask if causal_4d_mask is None else expanded_attn_mask + causal_4d_mask
+
+        return expanded_4d_mask
+
+    @staticmethod
+    def _make_causal_mask(
+        input_ids_shape: torch.Size,
+        dtype: torch.dtype,
+        device: torch.device,
+        past_key_values_length: int = 0,
+        sliding_window: Optional[int] = None,
+    ):
+        """
+        Make causal mask used for bi-directional self-attention.
+        """
+        bsz, tgt_len = input_ids_shape
+        mask = torch.full((tgt_len, tgt_len), torch.finfo(dtype).min, device=device)
+        mask_cond = torch.arange(mask.size(-1), device=device)
+        mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
+
+        mask = mask.to(dtype)
+
+        if past_key_values_length > 0:
+            mask = torch.cat([torch.zeros(tgt_len, past_key_values_length, dtype=dtype, device=device), mask], dim=-1)
+
+        # add lower triangular sliding window mask if necessary
+        if sliding_window is not None:
+            diagonal = past_key_values_length - sliding_window + 1
+
+            context_mask = 1 - torch.triu(torch.ones_like(mask, dtype=torch.int), diagonal=diagonal)
+            mask.masked_fill_(context_mask.bool(), torch.finfo(dtype).min)
+
+        return mask[None, None, :, :].expand(bsz, 1, tgt_len, tgt_len + past_key_values_length)
+
+    @staticmethod
+    def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
+        """
+        Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
+        """
+        bsz, src_len = mask.size()
+        tgt_len = tgt_len if tgt_len is not None else src_len
+
+        expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype)
+
+        inverted_mask = 1.0 - expanded_mask
+
+        return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)
+
+
+def _prepare_4d_causal_attention_mask(
+    attention_mask: Optional[torch.Tensor],
+    input_shape: Union[torch.Size, Tuple, List],
+    inputs_embeds: torch.Tensor,
+    past_key_values_length: int,
+    sliding_window: Optional[int] = None,
+):
+    """
+    Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape
+    `(batch_size, key_value_length)`
+
+    Args:
+        attention_mask (`torch.Tensor` or `None`):
+            A 2D attention mask of shape `(batch_size, key_value_length)`
+        input_shape (`tuple(int)` or `list(int)` or `torch.Size`):
+            The input shape should be a tuple that defines `(batch_size, query_length)`.
+        inputs_embeds (`torch.Tensor`):
+            The embedded inputs as a torch Tensor.
+        past_key_values_length (`int`):
+            The length of the key value cache.
+        sliding_window (`int`, *optional*):
+            If the model uses windowed attention, a sliding window should be passed.
+    """
+    attn_mask_converter = AttentionMaskConverter(is_causal=True, sliding_window=sliding_window)
+
+    key_value_length = input_shape[-1] + past_key_values_length
+
+    # 4d mask is passed through the layers
+    if attention_mask is not None:
+        attention_mask = attn_mask_converter.to_4d(
+            attention_mask, input_shape[-1], key_value_length, dtype=inputs_embeds.dtype
+        )
+    else:
+        attention_mask = attn_mask_converter.to_causal_4d(
+            input_shape[0], input_shape[-1], key_value_length, dtype=inputs_embeds.dtype, device=inputs_embeds.device
+        )
+
+    return attention_mask
+
+
+def _prepare_4d_attention_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
+    """
+    Creates a non-causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape
+    `(batch_size, key_value_length)`
+
+    Args:
+        mask (`torch.Tensor` or `None`):
+            A 2D attention mask of shape `(batch_size, key_value_length)`
+        dtype (`torch.dtype`):
+            The torch dtype the created mask shall have.
+        tgt_len (`int`):
+            The target length or query length the created mask shall have.
+    """
+    return AttentionMaskConverter._expand_mask(mask=mask, dtype=dtype, tgt_len=tgt_len)
+
+
+def _create_4d_causal_attention_mask(
+    input_shape: Union[torch.Size, Tuple, List],
+    dtype: torch.dtype,
+    device: torch.device,
+    past_key_values_length: int = 0,
+    sliding_window: Optional[int] = None,
+):
+    """
+    Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)`
+
+    Args:
+        input_shape (`tuple(int)` or `list(int)` or `torch.Size`):
+            The input shape should be a tuple that defines `(batch_size, query_length)`.
+        dtype (`torch.dtype`):
+            The torch dtype the created mask shall have.
+        device (`int`):
+            The torch device the created mask shall have.
+        sliding_window (`int`, *optional*):
+            If the model uses windowed attention, a sliding window should be passed.
+    """
+    attn_mask_converter = AttentionMaskConverter(is_causal=True, sliding_window=sliding_window)
+
+    key_value_length = past_key_values_length + input_shape[-1]
+    attention_mask = attn_mask_converter.to_causal_4d(
+        input_shape[0], input_shape[-1], key_value_length, dtype=dtype, device=device
+    )
+
+    return attention_mask

modeling_llama2.py

@@ -0,0 +1,837 @@
+import math
+import warnings
+from functools import partial
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.nn.functional as F
+import torch.utils.checkpoint
+from torch import nn
+from transformers.modeling_attn_mask_utils import _prepare_4d_causal_attention_mask, _prepare_4d_causal_attention_mask_for_sdpa
+
+
+import copy
+import os
+import sys
+
+dir_path = os.path.dirname(os.path.realpath(__file__))
+sys.path.insert(0, dir_path)
+
+import transformers
+from transformers.models.llama.modeling_llama import *
+
+def _get_unpad_data(attention_mask):
+    seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32)
+    indices = torch.nonzero(attention_mask.flatten(), as_tuple=False).flatten()
+    max_seqlen_in_batch = seqlens_in_batch.max().item()
+    cu_seqlens = F.pad(torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.torch.int32), (1, 0))
+    return (
+        indices,
+        cu_seqlens,
+        max_seqlen_in_batch,
+    )
+
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+from .modeling_attn_mask_utils import _prepare_4d_causal_attention_mask
+from .configuration_mplug_owl2 import LlamaConfig
+# from modeling_attn_mask_utils import _prepare_4d_causal_attention_mask
+# from configuration_mplug_owl2 import LlamaConfig
+
+class MultiwayNetwork(nn.Module):
+
+    def __init__(self, module_provider, num_multiway=2):
+        super(MultiwayNetwork, self).__init__()
+
+        self.multiway = torch.nn.ModuleList([module_provider() for _ in range(num_multiway)])
+    
+    def forward(self, hidden_states, multiway_indices):
+
+        if len(self.multiway) == 1:
+            return self.multiway[0](hidden_states)
+
+        output_hidden_states = torch.empty_like(hidden_states)
+        
+        for idx, subway in enumerate(self.multiway):
+            local_indices = multiway_indices.eq(idx).nonzero(as_tuple=True)
+            hidden = hidden_states[local_indices].unsqueeze(1).contiguous()
+            if hidden.numel():
+                output = subway(hidden)
+                if isinstance(output, tuple):
+                    output = output[0]
+                output = output.squeeze(1)
+                output_hidden_states[local_indices] = output
+        
+        return output_hidden_states.contiguous()
+    
+
+class LlamaAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: LlamaConfig, layer_idx: Optional[int] = None):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        if layer_idx is None:
+            logger.warning_once(
+                f"Instantiating {self.__class__.__name__} without passing `layer_idx` is not recommended and will "
+                "to errors during the forward call, if caching is used. Please make sure to provide a `layer_idx` "
+                "when creating this class."
+            )
+            
+        self.attention_dropout = config.attention_dropout
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        self.num_key_value_heads = config.num_key_value_heads
+        self.num_key_value_groups = self.num_heads // self.num_key_value_heads
+        self.max_position_embeddings = config.max_position_embeddings
+        self.rope_theta = config.rope_theta
+        self.is_causal = True
+
+        if (self.head_dim * self.num_heads) != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
+                f" and `num_heads`: {self.num_heads})."
+            )
+        self.q_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=config.attention_bias)
+        self.k_proj = MultiwayNetwork(module_provider=partial(
+            nn.Linear, in_features=self.hidden_size, out_features=self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
+        )
+        self.v_proj = MultiwayNetwork(module_provider=partial(
+            nn.Linear, in_features=self.hidden_size, out_features=self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
+        )
+        self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.hidden_size, bias=config.attention_bias)
+        self._init_rope()
+
+    def _init_rope(self):
+        if self.config.rope_scaling is None:
+            self.rotary_emb = LlamaRotaryEmbedding(
+                self.head_dim,
+                max_position_embeddings=self.max_position_embeddings,
+                base=self.rope_theta,
+            )
+        else:
+            scaling_type = self.config.rope_scaling["type"]
+            scaling_factor = self.config.rope_scaling["factor"]
+            if scaling_type == "linear":
+                self.rotary_emb = LlamaLinearScalingRotaryEmbedding(
+                    self.head_dim,
+                    max_position_embeddings=self.max_position_embeddings,
+                    scaling_factor=scaling_factor,
+                    base=self.rope_theta,
+                )
+            elif scaling_type == "dynamic":
+                self.rotary_emb = LlamaDynamicNTKScalingRotaryEmbedding(
+                    self.head_dim,
+                    max_position_embeddings=self.max_position_embeddings,
+                    scaling_factor=scaling_factor,
+                    base=self.rope_theta,
+                )
+            else:
+                raise ValueError(f"Unknown RoPE scaling type {scaling_type}")
+
+    def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
+        return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        modality_indicators: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        padding_mask: Optional[torch.LongTensor] = None,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states, )
+        key_states = self.k_proj(hidden_states, modality_indicators)
+        value_states = self.v_proj(hidden_states, modality_indicators)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            kv_seq_len += past_key_value[0].shape[-2]
+        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            # reuse k, v, self_attention
+            key_states = torch.cat([past_key_value[0], key_states], dim=2)
+            value_states = torch.cat([past_key_value[1], value_states], dim=2)
+
+        past_key_value = (key_states, value_states) if use_cache else None
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+
+        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
+            raise ValueError(
+                f"Attention weights should be of size {(bsz, self.num_heads, q_len, kv_seq_len)}, but is"
+                f" {attn_weights.size()}"
+            )
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+            attn_weights = attn_weights + attention_mask
+
+        # upcast attention to fp32
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query_states.dtype)
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+    
+
+class LlamaFlashAttention2(LlamaAttention):
+    """
+    Llama flash attention module. This module inherits from `LlamaAttention` as the weights of the module stays
+    untouched. The only required change would be on the forward pass where it needs to correctly call the public API of
+    flash attention and deal with padding tokens in case the input contains any of them.
+    """
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        # TODO: Should be removed once Flash Attention for RoCm is bumped to 2.1.
+        # flash_attn<2.1 generates top-left aligned causal mask, while what is needed here is bottom-right alignement, that was made default for flash_attn>=2.1. This attribute is used to handle this difference. Reference: https://github.com/Dao-AILab/flash-attention/releases/tag/v2.1.0.
+        # Beware that with flash_attn<2.1, using q_seqlen != k_seqlen (except for the case q_seqlen == 1) produces a wrong mask (top-left).
+        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        modality_indicators: torch.Tensor,
+        attention_mask: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        # LlamaFlashAttention2 attention does not support output_attentions
+        if "padding_mask" in kwargs:
+            warnings.warn(
+                "Passing `padding_mask` is deprecated and will be removed in v4.37. Please make sure use `attention_mask` instead.`"
+            )
+
+            # overwrite attention_mask with padding_mask
+            attention_mask = kwargs.pop("padding_mask")
+
+        output_attentions = False
+
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states, modality_indicators)
+        value_states = self.v_proj(hidden_states, modality_indicators)
+
+        # Flash attention requires the input to have the shape
+        # batch_size x seq_length x head_dim x hidden_dim
+        # therefore we just need to keep the original shape
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        # TODO: These transpose are quite inefficient but Flash Attention requires the layout [batch_size, sequence_length, num_heads, head_dim]. We would need to refactor the KV cache
+        # to be able to avoid many of these transpose/reshape/view.
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        dropout_rate = self.attention_dropout if self.training else 0.0
+
+        # In PEFT, usually we cast the layer norms in float32 for training stability reasons
+        # therefore the input hidden states gets silently casted in float32. Hence, we need
+        # cast them back in the correct dtype just to be sure everything works as expected.
+        # This might slowdown training & inference so it is recommended to not cast the LayerNorms
+        # in fp32. (LlamaRMSNorm handles it correctly)
+
+        input_dtype = query_states.dtype
+        if input_dtype == torch.float32:
+            if torch.is_autocast_enabled():
+                target_dtype = torch.get_autocast_gpu_dtype()
+            # Handle the case where the model is quantized
+            elif hasattr(self.config, "_pre_quantization_dtype"):
+                target_dtype = self.config._pre_quantization_dtype
+            else:
+                target_dtype = self.q_proj.weight.dtype
+
+            logger.warning_once(
+                f"The input hidden states seems to be silently casted in float32, this might be related to"
+                f" the fact you have upcasted embedding or layer norm layers in float32. We will cast back the input in"
+                f" {target_dtype}."
+            )
+
+            query_states = query_states.to(target_dtype)
+            key_states = key_states.to(target_dtype)
+            value_states = value_states.to(target_dtype)
+
+        attn_output = self._flash_attention_forward(
+            query_states, key_states, value_states, attention_mask, q_len, dropout=dropout_rate
+        )
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size).contiguous()
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+    def _flash_attention_forward(
+        self, query_states, key_states, value_states, attention_mask, query_length, dropout=0.0, softmax_scale=None
+    ):
+        """
+        Calls the forward method of Flash Attention - if the input hidden states contain at least one padding token
+        first unpad the input, then computes the attention scores and pad the final attention scores.
+
+        Args:
+            query_states (`torch.Tensor`):
+                Input query states to be passed to Flash Attention API
+            key_states (`torch.Tensor`):
+                Input key states to be passed to Flash Attention API
+            value_states (`torch.Tensor`):
+                Input value states to be passed to Flash Attention API
+            attention_mask (`torch.Tensor`):
+                The padding mask - corresponds to a tensor of size `(batch_size, seq_len)` where 0 stands for the
+                position of padding tokens and 1 for the position of non-padding tokens.
+            dropout (`int`, *optional*):
+                Attention dropout
+            softmax_scale (`float`, *optional*):
+                The scaling of QK^T before applying softmax. Default to 1 / sqrt(head_dim)
+        """
+        if not self._flash_attn_uses_top_left_mask:
+            causal = self.is_causal
+        else:
+            # TODO: Remove the `query_length != 1` check once Flash Attention for RoCm is bumped to 2.1. For details, please see the comment in LlamaFlashAttention2 __init__.
+            causal = self.is_causal and query_length != 1
+
+        # Contains at least one padding token in the sequence
+        if attention_mask is not None:
+            batch_size = query_states.shape[0]
+            query_states, key_states, value_states, indices_q, cu_seq_lens, max_seq_lens = self._upad_input(
+                query_states, key_states, value_states, attention_mask, query_length
+            )
+
+            cu_seqlens_q, cu_seqlens_k = cu_seq_lens
+            max_seqlen_in_batch_q, max_seqlen_in_batch_k = max_seq_lens
+
+            attn_output_unpad = flash_attn_varlen_func(
+                query_states,
+                key_states,
+                value_states,
+                cu_seqlens_q=cu_seqlens_q,
+                cu_seqlens_k=cu_seqlens_k,
+                max_seqlen_q=max_seqlen_in_batch_q,
+                max_seqlen_k=max_seqlen_in_batch_k,
+                dropout_p=dropout,
+                softmax_scale=softmax_scale,
+                causal=causal,
+            )
+
+            attn_output = pad_input(attn_output_unpad, indices_q, batch_size, query_length)
+        else:
+            attn_output = flash_attn_func(
+                query_states, key_states, value_states, dropout, softmax_scale=softmax_scale, causal=causal
+            )
+
+        return attn_output
+
+    def _upad_input(self, query_layer, key_layer, value_layer, attention_mask, query_length):
+        indices_k, cu_seqlens_k, max_seqlen_in_batch_k = _get_unpad_data(attention_mask)
+        batch_size, kv_seq_len, num_key_value_heads, head_dim = key_layer.shape
+
+        key_layer = index_first_axis(
+            key_layer.reshape(batch_size * kv_seq_len, num_key_value_heads, head_dim), indices_k
+        )
+        value_layer = index_first_axis(
+            value_layer.reshape(batch_size * kv_seq_len, num_key_value_heads, head_dim), indices_k
+        )
+        if query_length == kv_seq_len:
+            query_layer = index_first_axis(
+                query_layer.reshape(batch_size * kv_seq_len, self.num_heads, head_dim), indices_k
+            )
+            cu_seqlens_q = cu_seqlens_k
+            max_seqlen_in_batch_q = max_seqlen_in_batch_k
+            indices_q = indices_k
+        elif query_length == 1:
+            max_seqlen_in_batch_q = 1
+            cu_seqlens_q = torch.arange(
+                batch_size + 1, dtype=torch.int32, device=query_layer.device
+            )  # There is a memcpy here, that is very bad.
+            indices_q = cu_seqlens_q[:-1]
+            query_layer = query_layer.squeeze(1)
+        else:
+            # The -q_len: slice assumes left padding.
+            attention_mask = attention_mask[:, -query_length:]
+            query_layer, indices_q, cu_seqlens_q, max_seqlen_in_batch_q = unpad_input(query_layer, attention_mask)
+
+        return (
+            query_layer,
+            key_layer,
+            value_layer,
+            indices_q,
+            (cu_seqlens_q, cu_seqlens_k),
+            (max_seqlen_in_batch_q, max_seqlen_in_batch_k),
+        )
+
+
+class LlamaSdpaAttention(LlamaAttention):
+    """
+    Llama attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from
+    `LlamaAttention` as the weights of the module stays untouched. The only changes are on the forward pass to adapt to
+    SDPA API.
+    """
+
+    # Adapted from LlamaAttention.forward
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        modality_indicators: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        if output_attentions:
+            # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
+            logger.warning_once(
+                "LlamaModel is using LlamaSdpaAttention, but `torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to the manual attention implementation, "
+                'but specifying the manual implementation will be required from Transformers version v5.0.0 onwards. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
+            )
+            return super().forward(
+                hidden_states=hidden_states,
+                modality_indicators=modality_indicators,
+                attention_mask=attention_mask,
+                position_ids=position_ids,
+                past_key_value=past_key_value,
+                output_attentions=output_attentions,
+                use_cache=use_cache,
+            )
+
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states, modality_indicators)
+        value_states = self.v_proj(hidden_states, modality_indicators)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
+        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+
+        # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
+        # Reference: https://github.com/pytorch/pytorch/issues/112577.
+        if query_states.device.type == "cuda" and attention_mask is not None:
+            query_states = query_states.contiguous()
+            key_states = key_states.contiguous()
+            value_states = value_states.contiguous()
+
+        attn_output = torch.nn.functional.scaled_dot_product_attention(
+            query_states,
+            key_states,
+            value_states,
+            attn_mask=attention_mask,
+            dropout_p=self.attention_dropout if self.training else 0.0,
+            # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1.
+            is_causal=self.is_causal and attention_mask is None and q_len > 1,
+        )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        return attn_output, None, past_key_value
+
+
+
+LLAMA_ATTENTION_CLASSES = {
+    "eager": LlamaAttention,
+    "flash_attention_2": LlamaFlashAttention2,
+    "sdpa": LlamaSdpaAttention,
+}
+
+class LlamaDecoderLayer(nn.Module):
+    def __init__(self, config: LlamaConfig, layer_idx):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        self.self_attn = LlamaAttention(config=config)
+        self.self_attn = LLAMA_ATTENTION_CLASSES[config._attn_implementation](config=config, layer_idx=layer_idx)
+        self.mlp = LlamaMLP(config)
+        self.input_layernorm = MultiwayNetwork(module_provider=partial(
+            LlamaRMSNorm, hidden_size=config.hidden_size, eps=config.rms_norm_eps
+        ))
+        self.post_attention_layernorm = MultiwayNetwork(module_provider=partial(
+            LlamaRMSNorm, hidden_size=config.hidden_size, eps=config.rms_norm_eps
+        ))
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        modality_indicators: torch.Tensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`, *optional*): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+            use_cache (`bool`, *optional*):
+                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
+                (see `past_key_values`).
+            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+        """
+
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states, modality_indicators)
+
+        # Self Attention
+        hidden_states, self_attn_weights, present_key_value = self.self_attn(
+            hidden_states=hidden_states,
+            modality_indicators=modality_indicators,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_value=past_key_value,
+            output_attentions=output_attentions,
+            use_cache=use_cache,
+        )
+        hidden_states = residual + hidden_states
+
+        # Fully Connected
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states, modality_indicators)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + hidden_states
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (self_attn_weights,)
+
+        if use_cache:
+            outputs += (present_key_value,)
+
+        return outputs
+
+
+def model_forward(
+    self,
+    input_ids: torch.LongTensor = None,
+    modality_indicators: torch.Tensor = None,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_values: Optional[List[torch.FloatTensor]] = None,
+    inputs_embeds: 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,
+) -> Union[Tuple, BaseModelOutputWithPast]:
+    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
+
+    # retrieve input_ids and inputs_embeds
+    if input_ids is not None and inputs_embeds is not None:
+        raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
+    elif input_ids is not None:
+        batch_size, seq_length = input_ids.shape
+    elif inputs_embeds is not None:
+        batch_size, seq_length, _ = inputs_embeds.shape
+    else:
+        raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")
+    
+    seq_length_with_past = seq_length
+    past_key_values_length = 0
+
+    if past_key_values is not None:
+        past_key_values_length = past_key_values[0][0].shape[2]
+        seq_length_with_past = seq_length_with_past + past_key_values_length
+
+    if position_ids is None:
+        device = input_ids.device if input_ids is not None else inputs_embeds.device
+        position_ids = torch.arange(
+            past_key_values_length, seq_length + past_key_values_length, dtype=torch.long, device=device
+        )
+        position_ids = position_ids.unsqueeze(0).view(-1, seq_length)
+    else:
+        position_ids = position_ids.view(-1, seq_length).long()
+
+    if inputs_embeds is None:
+        inputs_embeds = self.embed_tokens(input_ids)
+    # embed positions
+    if attention_mask is None:
+        attention_mask = torch.ones(
+            (batch_size, seq_length_with_past), dtype=torch.bool, device=inputs_embeds.device
+        )
+    
+    if self._use_flash_attention_2:
+            # 2d mask is passed through the layers
+            attention_mask = attention_mask if (attention_mask is not None and 0 in attention_mask) else None
+    elif self._use_sdpa and not output_attentions:
+            # output_attentions=True can not be supported when using SDPA, and we fall back on
+            # the manual implementation that requires a 4D causal mask in all cases.
+            attention_mask = _prepare_4d_causal_attention_mask_for_sdpa(
+                attention_mask,
+                (batch_size, seq_length),
+                inputs_embeds,
+                past_key_values_length,
+            )
+    else:
+            # 4d mask is passed through the layers
+            attention_mask = _prepare_4d_causal_attention_mask(
+                attention_mask, (batch_size, seq_length), inputs_embeds, past_key_values_length
+            )
+
+    hidden_states = inputs_embeds
+
+    if self.gradient_checkpointing and self.training:
+        if use_cache:
+            logger.warning_once(
+                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+            )
+            use_cache = False
+
+    # decoder layers
+    all_hidden_states = () if output_hidden_states else None
+    all_self_attns = () if output_attentions else None
+    next_decoder_cache = () if use_cache else None
+
+    for idx, decoder_layer in enumerate(self.layers):
+        if output_hidden_states:
+            all_hidden_states += (hidden_states,)
+
+        past_key_value = past_key_values[idx] if past_key_values is not None else None
+
+        if self.gradient_checkpointing and self.training:
+
+            def create_custom_forward(module):
+                def custom_forward(*inputs):
+                    # None for past_key_value
+                    return module(*inputs, past_key_value, output_attentions)
+
+                return custom_forward
+
+            layer_outputs = torch.utils.checkpoint.checkpoint(
+                create_custom_forward(decoder_layer),
+                hidden_states,
+                modality_indicators,
+                attention_mask,
+                position_ids,
+            )
+        else:
+            layer_outputs = decoder_layer(
+                hidden_states,
+                modality_indicators=modality_indicators,
+                attention_mask=attention_mask,
+                position_ids=position_ids,
+                past_key_value=past_key_value,
+                output_attentions=output_attentions,
+                use_cache=use_cache,
+            )
+
+        hidden_states = layer_outputs[0]
+
+        if use_cache:
+            next_decoder_cache += (layer_outputs[2 if output_attentions else 1],)
+
+        if output_attentions:
+            all_self_attns += (layer_outputs[1],)
+
+    hidden_states = self.norm(hidden_states)
+
+    # add hidden states from the last decoder layer
+    if output_hidden_states:
+        all_hidden_states += (hidden_states,)
+
+    next_cache = next_decoder_cache if use_cache else None
+    if not return_dict:
+        return tuple(v for v in [hidden_states, next_cache, all_hidden_states, all_self_attns] if v is not None)
+    return BaseModelOutputWithPast(
+        last_hidden_state=hidden_states,
+        past_key_values=next_cache,
+        hidden_states=all_hidden_states,
+        attentions=all_self_attns,
+    )
+
+
+def causal_model_forward(
+    self,
+    input_ids: torch.LongTensor = None,
+    modality_indicators: torch.Tensor = None,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_values: Optional[List[torch.FloatTensor]] = None,
+    inputs_embeds: Optional[torch.FloatTensor] = None,
+    labels: Optional[torch.LongTensor] = None,
+    use_cache: Optional[bool] = None,
+    output_attentions: Optional[bool] = None,
+    output_hidden_states: Optional[bool] = None,
+    return_dict: Optional[bool] = None,
+) -> Union[Tuple, CausalLMOutputWithPast]:
+    r"""
+    Args:
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+    Returns:
+
+    Example:
+
+    ```python
+    >>> from transformers import AutoTokenizer, LlamaForCausalLM
+
+    >>> model = LlamaForCausalLM.from_pretrained(PATH_TO_CONVERTED_WEIGHTS)
+    >>> tokenizer = AutoTokenizer.from_pretrained(PATH_TO_CONVERTED_TOKENIZER)
+
+    >>> prompt = "Hey, are you conscious? Can you talk to me?"
+    >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+    >>> # Generate
+    >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+    >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+    "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+    ```"""
+
+    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
+    )
+    return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+    # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+    outputs = self.model(
+        input_ids=input_ids,
+        modality_indicators=modality_indicators,
+        attention_mask=attention_mask,
+        position_ids=position_ids,
+        past_key_values=past_key_values,
+        inputs_embeds=inputs_embeds,
+        use_cache=use_cache,
+        output_attentions=output_attentions,
+        output_hidden_states=output_hidden_states,
+        return_dict=return_dict,
+    )
+
+    hidden_states = outputs[0]
+    if self.config.pretraining_tp > 1:
+        lm_head_slices = self.lm_head.weight.split(self.vocab_size // self.config.pretraining_tp, dim=0)
+        logits = [F.linear(hidden_states, lm_head_slices[i]) for i in range(self.config.pretraining_tp)]
+        logits = torch.cat(logits, dim=-1)
+    else:
+        logits = self.lm_head(hidden_states)
+    logits = logits.float()
+
+    loss = None
+    if labels is not None:
+        # Shift so that tokens < n predict n
+        shift_logits = logits[..., :-1, :].contiguous()
+        shift_labels = labels[..., 1:].contiguous()
+        # Flatten the tokens
+        loss_fct = CrossEntropyLoss()
+        shift_logits = shift_logits.view(-1, self.config.vocab_size)
+        shift_labels = shift_labels.view(-1)
+        # Enable model parallelism
+        shift_labels = shift_labels.to(shift_logits.device)
+        loss = loss_fct(shift_logits, shift_labels)
+
+    if not return_dict:
+        output = (logits,) + outputs[1:]
+        return (loss,) + output if loss is not None else output
+
+    return CausalLMOutputWithPast(
+        loss=loss,
+        logits=logits,
+        past_key_values=outputs.past_key_values,
+        hidden_states=outputs.hidden_states,
+        attentions=outputs.attentions,
+    )
+
+def replace_llama_modality_adaptive():
+    transformers.models.llama.configuration_llama.LlamaConfig = LlamaConfig
+    transformers.models.llama.modeling_llama.LlamaAttention = LlamaAttention
+    transformers.models.llama.modeling_llama.LlamaFlashAttention2 = LlamaFlashAttention2
+    transformers.models.llama.modeling_llama.LlamaSdpaAttention = LlamaSdpaAttention
+    transformers.models.llama.modeling_llama.LlamaDecoderLayer = LlamaDecoderLayer
+    transformers.models.llama.modeling_llama.LlamaModel.forward = model_forward
+    transformers.models.llama.modeling_llama.LlamaForCausalLM.forward = causal_model_forward
+
+    
+if __name__ == "__main__":
+    replace_llama_modality_adaptive()
+    config = transformers.LlamaConfig.from_pretrained('/cpfs01/shared/public/test/vicuna-7b-v1.5/')
+    model = transformers.LlamaForCausalLM(config)
+    print(model)

modeling_mplug_owl2.py

@@ -0,0 +1,469 @@
+#    Copyright 2023 Haotian Liu & Qinghao Ye (Modified from LLaVA)
+#
+#    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 abc import ABC, abstractmethod
+from typing import List, Optional, Tuple, Union
+from datasets import load_dataset
+import torch
+import torch.nn as nn
+from torch.nn import CrossEntropyLoss
+import numpy as np
+import copy
+import os
+import sys
+from PIL import Image
+import requests
+from io import BytesIO
+
+dir_path = os.path.dirname(os.path.realpath(__file__))
+sys.path.insert(0, dir_path)
+
+from transformers import AutoConfig, AutoModelForCausalLM, AutoTokenizer, CLIPImageProcessor, LlamaConfig, LlamaModel, LlamaForCausalLM
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
+from configuration_mplug_owl2 import MPLUGOwl2Config, MplugOwlVisionConfig, MplugOwlVisualAbstractorConfig
+from visual_encoder import MplugOwlVisionModel, MplugOwlVisualAbstractorModel
+from modeling_llama2 import replace_llama_modality_adaptive
+IGNORE_INDEX = -100
+IMAGE_TOKEN_INDEX = -200
+DEFAULT_IMAGE_TOKEN = "<|image|>"
+from icecream import ic
+
+def tokenizer_image_token(prompt, tokenizer, image_token_index=IMAGE_TOKEN_INDEX, return_tensors=None):
+    prompt_chunks = [tokenizer(chunk).input_ids if len(chunk) > 0 else [] for chunk in prompt.split(DEFAULT_IMAGE_TOKEN)]
+
+    def insert_separator(X, sep):
+        return [ele for sublist in zip(X, [sep]*len(X)) for ele in sublist][:-1]
+
+    input_ids = []
+    offset = 0
+    if len(prompt_chunks) > 0 and len(prompt_chunks[0]) > 0 and prompt_chunks[0][0] == tokenizer.bos_token_id:
+        offset = 1
+        input_ids.append(prompt_chunks[0][0])
+
+    for x in insert_separator(prompt_chunks, [image_token_index] * (offset + 1)):
+        input_ids.extend(x[offset:])
+
+    if return_tensors is not None:
+        if return_tensors == 'pt':
+            return torch.tensor(input_ids, dtype=torch.long)
+        raise ValueError(f'Unsupported tensor type: {return_tensors}')
+    return input_ids
+
+def expand2square(pil_img, background_color):
+        from PIL import Image
+        width, height = pil_img.size
+        if width == height:
+            return pil_img
+        elif width > height:
+            result = Image.new(pil_img.mode, (width, width), background_color)
+            result.paste(pil_img, (0, (width - height) // 2))
+            return result
+        else:
+            result = Image.new(pil_img.mode, (height, height), background_color)
+            result.paste(pil_img, ((height - width) // 2, 0))
+            return result
+
+def norm_cdf(x):
+    return 0.5 * (1 + torch.erf(x / torch.sqrt(torch.tensor(2.0))))
+
+def optimize_score_map_pytorch_cuda(c, seed=0, original_seed=20020, num_iterations=100):
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    
+    torch.manual_seed(seed)
+    np.random.seed(seed)
+
+    c = torch.tensor(c, dtype=torch.float32, device=device, requires_grad=False)
+    initial_scores = torch.rand(c.shape[0], device=device, requires_grad=True)
+    
+    optimizer = torch.optim.Adam([initial_scores], lr=0.1)
+
+    for _ in range(num_iterations):
+        optimizer.zero_grad()
+        sum_log_diff = torch.sum(c * torch.log(torch.maximum(torch.sigmoid(initial_scores[:, None] - initial_scores), torch.tensor(1e-6, device=device))))
+        sum_squares = torch.sum(initial_scores ** 2) / 2
+
+        loss = -(sum_log_diff - sum_squares)
+        loss.backward()
+        optimizer.step()
+    
+    optimized_scores = initial_scores.detach().cpu().numpy()
+    min_score, max_score = np.min(optimized_scores), np.max(optimized_scores)
+    
+    # Scale scores to 0-100
+    scaled_scores = 100 * (optimized_scores - min_score) / (max_score - min_score)
+    
+    # Reset the seed
+    np.random.seed(original_seed)
+    return scaled_scores[-1]
+
+def softmax(logits):
+    # exp_logits = np.exp(logits - np.max(logits))
+    probs = np.exp(logits) / np.sum(np.exp(logits))
+    return probs
+    # return exp_logits / exp_logits.sum()
+
+def update_matrix(anchor_matrix, scores, indices):
+    n = anchor_matrix.shape[0]
+    new_row = np.zeros((1, n))
+    new_col = np.zeros((n + 1, 1))
+    new_row[0, indices] = scores
+    new_col[indices, 0] = 1-scores  # Assuming symmetric preference for simplicity
+    anchor_matrix = np.vstack([anchor_matrix, new_row])
+    anchor_matrix = np.hstack([anchor_matrix, new_col])
+
+    return anchor_matrix
+    
+
+class MPLUGOwl2MetaModel:
+    def __init__(self, config):
+        super(MPLUGOwl2MetaModel, self).__init__(config)
+        self.vision_model = MplugOwlVisionModel(
+            MplugOwlVisionConfig(**config.visual_config["visual_model"])
+        )
+        self.visual_abstractor = MplugOwlVisualAbstractorModel(
+            MplugOwlVisualAbstractorConfig(**config.visual_config["visual_abstractor"]), config.hidden_size
+        )
+    
+    def get_vision_tower(self):
+        vision_model = getattr(self, 'vision_model', None)
+        if type(vision_model) is list:
+            vision_model = vision_model[0]
+        return vision_model
+
+    def get_visual_abstractor(self):
+        visual_abstractor = getattr(self, 'visual_abstractor', None)
+        if type(visual_abstractor) is list:
+            visual_abstractor = visual_abstractor[0]
+        return visual_abstractor
+
+
+class MPLUGOwl2MetaForCausalLM(ABC):
+    @abstractmethod
+    def get_model(self):
+        pass
+
+    def encode_images(self, images):
+        image_features = self.get_model().vision_model(images).last_hidden_state
+        image_features = self.get_model().visual_abstractor(encoder_hidden_states=image_features).last_hidden_state
+        return image_features
+
+    def prepare_inputs_labels_for_multimodal(
+        self, input_ids, attention_mask, past_key_values, labels, images
+    ):
+        if images is None or input_ids.shape[1] == 1:
+            if past_key_values is not None and images is not None and input_ids.shape[1] == 1:
+                attention_mask = torch.ones((attention_mask.shape[0], past_key_values[-1][-1].shape[-2] + 1), dtype=attention_mask.dtype, device=attention_mask.device)
+            multiway_indices = torch.zeros_like(input_ids).long().to(self.device)
+            return input_ids, multiway_indices, attention_mask, past_key_values, None, labels
+        
+        if type(images) is list or images.ndim == 5:
+            concat_images = torch.cat([image for image in images], dim=0)
+            image_features = self.encode_images(concat_images)
+            split_sizes = [image.shape[0] for image in images]
+            image_features = torch.split(image_features, split_sizes, dim=0)
+            image_features = [x.flatten(0, 1) for x in image_features]
+        else:
+            image_features = self.encode_images(images)
+
+        new_input_embeds = []
+        new_modality_indicators = []
+        new_labels = [] if labels is not None else None
+        cur_image_idx = 0
+        for batch_idx, cur_input_ids in enumerate(input_ids):
+            if (cur_input_ids == IMAGE_TOKEN_INDEX).sum() == 0:
+                # multimodal LLM, but the current sample is not multimodal
+                # FIXME: this is a hacky fix, for deepspeed zero3 to work
+                half_len = cur_input_ids.shape[0] // 2
+                cur_image_features = image_features[cur_image_idx]
+                cur_input_embeds_1 = self.get_model().embed_tokens(cur_input_ids[:half_len])
+                cur_input_embeds_2 = self.get_model().embed_tokens(cur_input_ids[half_len:])
+                cur_input_embeds = torch.cat([cur_input_embeds_1, cur_image_features[0:0], cur_input_embeds_2], dim=0)
+                new_input_embeds.append(cur_input_embeds)
+                
+                cur_modality_indicators = torch.zeros(len(cur_input_embeds)).long().to(self.device)
+                new_modality_indicators.append(cur_modality_indicators)
+                if labels is not None:
+                    new_labels.append(labels[batch_idx])
+                cur_image_idx += 1
+                continue
+            image_token_indices = torch.where(cur_input_ids == IMAGE_TOKEN_INDEX)[0]
+            cur_new_input_embeds = []
+            cur_modality_indicators = []
+            if labels is not None:
+                cur_labels = labels[batch_idx]
+                cur_new_labels = []
+                assert cur_labels.shape == cur_input_ids.shape
+            while image_token_indices.numel() > 0:
+                # print("cur_image_idx", cur_image_idx)
+                cur_image_features = image_features[cur_image_idx]
+                image_token_start = image_token_indices[0]
+                cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids[:image_token_start]))
+                cur_new_input_embeds.append(cur_image_features)
+                
+                # Add modality indicator
+                assert image_token_start == len(cur_input_ids[:image_token_start])
+                cur_modality_indicators.append(torch.zeros(len(cur_input_ids[:image_token_start])).long())
+                cur_modality_indicators.append(torch.ones(len(cur_image_features)).long())
+                
+                if labels is not None:
+                    cur_new_labels.append(cur_labels[:image_token_start])
+                    cur_new_labels.append(torch.full((cur_image_features.shape[0],), IGNORE_INDEX, device=labels.device, dtype=labels.dtype))
+                    cur_labels = cur_labels[image_token_start+1:]
+                cur_image_idx += 1
+                cur_input_ids = cur_input_ids[image_token_start+1:]
+                image_token_indices = torch.where(cur_input_ids == IMAGE_TOKEN_INDEX)[0]
+            if cur_input_ids.numel() > 0:
+                cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids))
+                cur_modality_indicators.append(torch.zeros(len(cur_input_ids)).long())
+                if labels is not None:
+                    cur_new_labels.append(cur_labels)
+            cur_new_input_embeds = [x.to(device=self.device) for x in cur_new_input_embeds]
+            cur_new_input_embeds = torch.cat(cur_new_input_embeds, dim=0)
+            new_input_embeds.append(cur_new_input_embeds)
+            
+            # Modality
+            cur_modality_indicators = [x.to(device=self.device) for x in cur_modality_indicators]
+            cur_modality_indicators = torch.cat(cur_modality_indicators, dim=0)
+            new_modality_indicators.append(cur_modality_indicators)
+            
+            
+            if labels is not None:
+                cur_new_labels = torch.cat(cur_new_labels, dim=0)
+                new_labels.append(cur_new_labels)
+
+        if any(x.shape != new_input_embeds[0].shape for x in new_input_embeds):
+            max_len = max(x.shape[0] for x in new_input_embeds)
+            
+            # Embedding
+            new_input_embeds_align = []
+            for cur_new_embed in new_input_embeds:
+                cur_new_embed = torch.cat((cur_new_embed, torch.zeros((max_len - cur_new_embed.shape[0], cur_new_embed.shape[1]), dtype=cur_new_embed.dtype, device=cur_new_embed.device)), dim=0)
+                new_input_embeds_align.append(cur_new_embed)
+            new_input_embeds = torch.stack(new_input_embeds_align, dim=0)
+            
+            # Modality
+            new_modality_indicators_align = []
+            for cur_modality_indicator in new_modality_indicators:
+                cur_new_embed = torch.cat((cur_modality_indicator, torch.zeros(max_len - cur_modality_indicator.shape[0], dtype=cur_modality_indicator.dtype, device=cur_modality_indicator.device)), dim=0)
+                new_modality_indicators_align.append(cur_new_embed)
+            new_modality_indicators = torch.stack(new_modality_indicators_align, dim=0)
+            
+            # Label
+            if labels is not None:
+                new_labels_align = []
+                _new_labels = new_labels
+                for cur_new_label in new_labels:
+                    cur_new_label = torch.cat((cur_new_label, torch.full((max_len - cur_new_label.shape[0],), IGNORE_INDEX, dtype=cur_new_label.dtype, device=cur_new_label.device)), dim=0)
+                    new_labels_align.append(cur_new_label)
+                new_labels = torch.stack(new_labels_align, dim=0)
+            
+            # Attention Mask
+            if attention_mask is not None:
+                new_attention_mask = []
+                for cur_attention_mask, cur_new_labels, cur_new_labels_align in zip(attention_mask, _new_labels, new_labels):
+                    new_attn_mask_pad_left = torch.full((cur_new_labels.shape[0] - labels.shape[1],), True, dtype=attention_mask.dtype, device=attention_mask.device)
+                    new_attn_mask_pad_right = torch.full((cur_new_labels_align.shape[0] - cur_new_labels.shape[0],), False, dtype=attention_mask.dtype, device=attention_mask.device)
+                    cur_new_attention_mask = torch.cat((new_attn_mask_pad_left, cur_attention_mask, new_attn_mask_pad_right), dim=0)
+                    new_attention_mask.append(cur_new_attention_mask)
+                attention_mask = torch.stack(new_attention_mask, dim=0)
+                assert attention_mask.shape == new_labels.shape
+        else:
+            new_input_embeds = torch.stack(new_input_embeds, dim=0)
+            new_modality_indicators = torch.stack(new_modality_indicators, dim=0)
+            if labels is not None:
+                new_labels = torch.stack(new_labels, dim=0)
+
+            if attention_mask is not None:
+                new_attn_mask_pad_left = torch.full((attention_mask.shape[0], new_input_embeds.shape[1] - input_ids.shape[1]), True, dtype=attention_mask.dtype, device=attention_mask.device)
+                attention_mask = torch.cat((new_attn_mask_pad_left, attention_mask), dim=1)
+                assert attention_mask.shape == new_input_embeds.shape[:2]
+        return None, new_modality_indicators, attention_mask, past_key_values, new_input_embeds, new_labels
+
+
+
+class MPLUGOwl2LlamaModel(MPLUGOwl2MetaModel, LlamaModel):
+    config_class = MPLUGOwl2Config
+
+    def __init__(self, config: MPLUGOwl2Config):
+        super(MPLUGOwl2LlamaModel, self).__init__(config)
+
+
+class MPLUGOwl2LlamaForCausalLM(LlamaForCausalLM, MPLUGOwl2MetaForCausalLM):
+    config_class = MPLUGOwl2Config
+
+    def __init__(self, config):
+        super(LlamaForCausalLM, self).__init__(config)
+        self.model = MPLUGOwl2LlamaModel(config)
+        self.tokenizer = AutoTokenizer.from_pretrained("VQA-CityU/Compare2Score_1")
+        self.image_processor = CLIPImageProcessor.from_pretrained("VQA-CityU/Compare2Score_1")
+
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+        self.preferential_ids_ = [id_[1] for id_ in self.tokenizer(["inferior", "worse", "similar", "better", "superior"])["input_ids"]]
+        self.anchor_images = load_dataset("VQA-CityU/Anchor_images")
+        
+        self.weight_tensor = np.array([0., 0.25, 0.5, 0.75, 1.], dtype=np.float16)
+        self.anchor_matrix = np.array(
+            [[5.0000000e-01, 2.5912809e-01, 3.3130276e-04, 1.6087297e-06, 1.1803027e-09],
+             [7.4087191e-01, 5.0000000e-01, 2.4985345e-01, 9.9954158e-02, 1.8675303e-08],
+             [9.9966872e-01, 7.5014657e-01, 5.0000000e-01, 4.9968880e-01, 2.4852838e-01],
+             [9.9999839e-01, 9.0004587e-01, 5.0031120e-01, 5.0000000e-01, 2.5400183e-01],
+             [1.0000000e+00, 1.0000000e+00, 7.5147164e-01, 7.4599814e-01, 5.0000000e-01]], 
+            dtype=np.float32)
+        anchor_intervals = 5#16
+        num_anchor_image_per_interval = 1
+        num_anchor_image = anchor_intervals * num_anchor_image_per_interval
+        self.anchor_indices = np.arange(0,num_anchor_image)
+        # Initialize weights and apply final processing
+        self.post_init()
+        
+
+    def get_model(self):
+        return self.model
+    
+    def download_image(self, url):
+        response = requests.get(url)
+        return Image.open(BytesIO(response.content)).convert('RGB')
+
+    def load_image(self, path):
+        if path.startswith('http://') or path.startswith('https://'):
+            return self.download_image(path)
+        return Image.open(path).convert('RGB')
+    
+    def score(self, image_path):
+        prompt = "USER: <|image|> <|image|> Compared with the first image, what is your quality rating for second image? \nASSISTANT: The quality of the second image is"
+        input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to(self.device)
+        
+        anchor_images = [item['image'] for item in self.anchor_images['train']]
+        
+        probabilities = []
+        for index in self.anchor_indices:
+            anchor_image = anchor_images[index]
+            image = self.load_image(image_path)
+            images = [anchor_image, image]
+            images = [expand2square(img, tuple(int(x*255) for x in self.image_processor.image_mean)) for img in images]
+            image_tensor = self.image_processor.preprocess(images, return_tensors='pt')['pixel_values'].half().to(self.device)
+            
+            with torch.inference_mode():
+                output_logits = model(input_ids, images=image_tensor)["logits"][:, -1, self.preferential_ids_]
+                output_logits = output_logits.cpu().detach().numpy() / 100
+                print(output_logits)
+                probabilities.append(np.dot(softmax(output_logits),  self.weight_tensor))
+        updated_matrix = update_matrix(self.anchor_matrix, np.squeeze(np.array(probabilities)), self.anchor_indices)
+        score = optimize_score_map_pytorch_cuda(updated_matrix, seed=0, original_seed=20020, num_iterations=100)
+        print(score)
+        return score
+
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        # modality_indicators: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        images: Optional[torch.FloatTensor] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+        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
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+        input_ids, modality_indicators, attention_mask, past_key_values, inputs_embeds, labels = \
+            self.prepare_inputs_labels_for_multimodal(input_ids, attention_mask, past_key_values, labels, images)
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs = self.model(
+            input_ids=input_ids,
+            modality_indicators=modality_indicators,
+            attention_mask=attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict
+        )
+
+        hidden_states = outputs[0]
+        logits = self.lm_head(hidden_states)
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model/pipeline parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    def prepare_inputs_for_generation(
+        self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, **kwargs
+    ):
+        if past_key_values:
+            input_ids = input_ids[:, -1:]
+
+        # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
+        if inputs_embeds is not None and past_key_values is None:
+            model_inputs = {"inputs_embeds": inputs_embeds}
+        else:
+            model_inputs = {"input_ids": input_ids}
+
+        model_inputs.update(
+            {
+                "past_key_values": past_key_values,
+                "use_cache": kwargs.get("use_cache"),
+                "attention_mask": attention_mask,
+                "images": kwargs.get("images", None),
+            }
+        )
+        return model_inputs
+
+AutoConfig.register("mplug_owl2", MPLUGOwl2Config)
+AutoModelForCausalLM.register(MPLUGOwl2Config, MPLUGOwl2LlamaForCausalLM)
+
+replace_llama_modality_adaptive()
+
+if __name__ == "__main__":
+    # config = MPLUGOwl2Config.from_pretrained('VQA-CityU/Compare2Score_1')
+    from icecream import ic
+    # config = MPLUGOwl2Config()
+    # model =  AutoModelForCausalLM(config)
+    model = AutoModelForCausalLM.from_pretrained('VQA-CityU/Compare2Score_1', trust_remote_code=True, 
+                                             torch_dtype=torch.float16, device_map="auto")
+    
+    model.score("/home/zhw/IQA/code/NeurIPS24/Q-Align/playground/data/TID2013/distorted_images/i01_01_5.bmp")
+    url = "https://raw.githubusercontent.com/Q-Future/Q-Align/main/fig/singapore_flyer.jpg"
+    model.score(url)
+ 

visual_encoder.py

@@ -0,0 +1,922 @@
+import math
+from typing import Any, Optional, Tuple, Union
+
+from transformers.modeling_outputs import BaseModelOutput, BaseModelOutputWithPooling, BaseModelOutputWithPastAndCrossAttentions
+from transformers.modeling_utils import PreTrainedModel
+from transformers.pytorch_utils import find_pruneable_heads_and_indices, prune_linear_layer
+
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.utils.checkpoint
+from icecream import ic
+
+def get_abs_pos(abs_pos, tgt_size):
+    # abs_pos: L, C
+    # tgt_size: M
+    # return: M, C
+    src_size = int(math.sqrt(abs_pos.size(0)))
+    tgt_size = int(math.sqrt(tgt_size))
+    dtype = abs_pos.dtype
+
+    if src_size != tgt_size:
+        return F.interpolate(
+            abs_pos.float().reshape(1, src_size, src_size, -1).permute(0, 3, 1, 2),
+            size=(tgt_size, tgt_size),
+            mode="bicubic",
+            align_corners=False,
+        ).permute(0, 2, 3, 1).flatten(0, 2).to(dtype=dtype)
+    else:
+        return abs_pos
+
+# https://github.com/facebookresearch/mae/blob/efb2a8062c206524e35e47d04501ed4f544c0ae8/util/pos_embed.py#L20
+def get_2d_sincos_pos_embed(embed_dim, grid_size, cls_token=False):
+    """
+    grid_size: int of the grid height and width
+    return:
+    pos_embed: [grid_size*grid_size, embed_dim] or [1+grid_size*grid_size, embed_dim] (w/ or w/o cls_token)
+    """
+    grid_h = np.arange(grid_size, dtype=np.float32)
+    grid_w = np.arange(grid_size, dtype=np.float32)
+    grid = np.meshgrid(grid_w, grid_h)  # here w goes first
+    grid = np.stack(grid, axis=0)
+
+    grid = grid.reshape([2, 1, grid_size, grid_size])
+    pos_embed = get_2d_sincos_pos_embed_from_grid(embed_dim, grid)
+    if cls_token:
+        pos_embed = np.concatenate([np.zeros([1, embed_dim]), pos_embed], axis=0)
+    return pos_embed
+
+
+def get_2d_sincos_pos_embed_from_grid(embed_dim, grid):
+    assert embed_dim % 2 == 0
+
+    # use half of dimensions to encode grid_h
+    emb_h = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[0])  # (H*W, D/2)
+    emb_w = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[1])  # (H*W, D/2)
+
+    emb = np.concatenate([emb_h, emb_w], axis=1) # (H*W, D)
+    return emb
+
+
+def get_1d_sincos_pos_embed_from_grid(embed_dim, pos):
+    """
+    embed_dim: output dimension for each position
+    pos: a list of positions to be encoded: size (M,)
+    out: (M, D)
+    """
+    assert embed_dim % 2 == 0
+    omega = np.arange(embed_dim // 2, dtype=np.float32)
+    omega /= embed_dim / 2.
+    omega = 1. / 10000**omega  # (D/2,)
+
+    pos = pos.reshape(-1)  # (M,)
+    out = np.einsum('m,d->md', pos, omega)  # (M, D/2), outer product
+
+    emb_sin = np.sin(out) # (M, D/2)
+    emb_cos = np.cos(out) # (M, D/2)
+
+    emb = np.concatenate([emb_sin, emb_cos], axis=1)  # (M, D)
+    return emb
+
+
+
+class MplugOwlVisionEmbeddings(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.image_size = config.image_size
+        self.patch_size = config.patch_size
+
+        self.cls_token = nn.Parameter(torch.randn(1, 1, self.hidden_size))
+
+        self.patch_embed = nn.Conv2d(
+            in_channels=3,
+            out_channels=self.hidden_size,
+            kernel_size=self.patch_size,
+            stride=self.patch_size,
+            bias=False,
+        )
+
+        self.num_patches = (self.image_size // self.patch_size) ** 2
+
+        self.position_embedding = nn.Parameter(torch.randn(1, self.num_patches + 1, self.hidden_size))
+
+        self.pre_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+
+    def forward(self, pixel_values: torch.FloatTensor) -> torch.Tensor:
+        batch_size = pixel_values.size(0)
+        image_embeds = self.patch_embed(pixel_values)
+        image_embeds = image_embeds.flatten(2).transpose(1, 2)
+
+        class_embeds = self.cls_token.expand(batch_size, 1, -1).to(image_embeds.dtype)
+        embeddings = torch.cat([class_embeds, image_embeds], dim=1)
+        embeddings = embeddings + self.position_embedding[:, : embeddings.size(1)].to(image_embeds.dtype)
+        embeddings = self.pre_layernorm(embeddings)
+        return embeddings
+
+
+
+class MplugOwlVisionAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        if self.head_dim * self.num_heads != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size} and `num_heads`:"
+                f" {self.num_heads})."
+            )
+        self.scale = self.head_dim**-0.5
+        self.dropout = nn.Dropout(config.attention_dropout)
+
+        self.query_key_value = nn.Linear(self.hidden_size, 3 * self.hidden_size)
+        self.dense = nn.Linear(self.hidden_size, self.hidden_size)
+
+    def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
+        return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        head_mask: Optional[torch.Tensor] = None,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        """Input shape: Batch x Time x Channel"""
+
+        bsz, seq_len, embed_dim = hidden_states.size()
+
+        mixed_qkv = self.query_key_value(hidden_states)
+
+        mixed_qkv = mixed_qkv.reshape(bsz, seq_len, self.num_heads, 3, embed_dim // self.num_heads).permute(
+            3, 0, 2, 1, 4
+        )  # [3, b, np, sq, hn]
+        query_states, key_states, value_states = (
+            mixed_qkv[0],
+            mixed_qkv[1],
+            mixed_qkv[2],
+        )
+        # if self.config.use_flash_attn and flash_attn_func is not None:
+        if False:
+            # [b*sq, np, hn]
+            query_states = query_states.permute(0, 2, 1, 3).contiguous()
+            query_states = query_states.view(query_states.size(0) * query_states.size(1), query_states.size(2), -1)
+
+            key_states = key_states.permute(0, 2, 1, 3).contiguous()
+            key_states = key_states.view(key_states.size(0) * key_states.size(1), key_states.size(2), -1)
+
+            value_states = value_states.permute(0, 2, 1, 3).contiguous()
+            value_states = value_states.view(value_states.size(0) * value_states.size(1), value_states.size(2), -1)
+
+            cu_seqlens = torch.arange(
+                0, (bsz + 1) * seq_len, step=seq_len, dtype=torch.int32, device=query_states.device
+            )
+
+            context_layer = flash_attn_func(
+                query_states,
+                key_states,
+                value_states,
+                cu_seqlens,
+                cu_seqlens,
+                seq_len,
+                seq_len,
+                self.dropout if self.training else 0.0,
+                softmax_scale=self.scale,
+                causal=False,
+                return_attn_probs=False,
+            )
+            # [b*sq, np, hn] => [b, sq, np, hn]
+            context_layer = context_layer.view(bsz, seq_len, context_layer.size(1), context_layer.size(2))
+        else:
+            # Take the dot product between "query" and "key" to get the raw attention scores.
+            attention_scores = torch.matmul(query_states, key_states.transpose(-1, -2))
+
+            attention_scores = attention_scores * self.scale
+
+            # Normalize the attention scores to probabilities.
+            attention_probs = torch.softmax(attention_scores, dim=-1)
+
+            # This is actually dropping out entire tokens to attend to, which might
+            # seem a bit unusual, but is taken from the original Transformer paper.
+            attention_probs = self.dropout(attention_probs)
+
+            # Mask heads if we want to
+            if head_mask is not None:
+                attention_probs = attention_probs * head_mask
+
+            context_layer = torch.matmul(attention_probs, value_states).permute(0, 2, 1, 3)
+
+        new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size,)
+        context_layer = context_layer.reshape(new_context_layer_shape)
+
+        output = self.dense(context_layer)
+
+        outputs = (output, attention_probs) if output_attentions else (output, None)
+
+        return outputs
+
+
+class QuickGELU(nn.Module):
+    def forward(self, x: torch.Tensor):
+        return x * torch.sigmoid(1.702 * x)
+
+
+class MplugOwlMLP(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.activation_fn = QuickGELU()
+        self.fc1 = nn.Linear(config.hidden_size, config.intermediate_size)
+        self.fc2 = nn.Linear(config.intermediate_size, config.hidden_size)
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.fc1(hidden_states)
+        hidden_states = self.activation_fn(hidden_states)
+        hidden_states = self.fc2(hidden_states)
+        return hidden_states
+
+
+class MplugOwlVisionEncoderLayer(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        self.self_attn = MplugOwlVisionAttention(config)
+        self.input_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+        self.mlp = MplugOwlMLP(config)
+        self.post_attention_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: torch.Tensor,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.FloatTensor]:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+                `(config.encoder_attention_heads,)`.
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+        """
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states)
+        hidden_states, attn_weights = self.self_attn(
+            hidden_states=hidden_states,
+            head_mask=attention_mask,
+            output_attentions=output_attentions,
+        )
+        hidden_states = hidden_states + residual
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+
+        hidden_states = hidden_states + residual
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (attn_weights,)
+
+        return outputs
+    
+    
+class MplugOwlVisionEncoder(nn.Module):
+    """
+    Transformer encoder consisting of `config.num_hidden_layers` self attention layers. Each layer is a
+    [`MplugOwlVisionEncoderLayer`].
+
+    Args:
+        config (`MplugOwlVisionConfig`):
+            The corresponding vision configuration for the `MplugOwlEncoder`.
+    """
+
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.layers = nn.ModuleList([MplugOwlVisionEncoderLayer(config) for _ in range(config.num_hidden_layers)])
+        self.gradient_checkpointing = True
+
+    def forward(
+        self,
+        inputs_embeds,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, BaseModelOutput]:
+        r"""
+        Args:
+            inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
+                Embedded representation of the inputs. Should be float, not int tokens.
+            attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
+                Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
+
+                - 1 for tokens that are **not masked**,
+                - 0 for tokens that are **masked**.
+
+                [What are attention masks?](../glossary#attention-mask)
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+            output_hidden_states (`bool`, *optional*):
+                Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
+                for more detail.
+            return_dict (`bool`, *optional*):
+                Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
+        """
+        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
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        encoder_states = () if output_hidden_states else None
+        all_attentions = () if output_attentions else None
+
+        hidden_states = inputs_embeds
+        for idx, encoder_layer in enumerate(self.layers):
+            if output_hidden_states:
+                encoder_states = encoder_states + (hidden_states,)
+            if self.gradient_checkpointing and self.training:
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        return module(*inputs, output_attentions)
+
+                    return custom_forward
+
+                layer_outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(encoder_layer),
+                    hidden_states,
+                    attention_mask,
+                )
+            else:
+                layer_outputs = encoder_layer(
+                    hidden_states,
+                    attention_mask,
+                    output_attentions=output_attentions,
+                )
+
+            hidden_states = layer_outputs[0]
+
+            if output_attentions:
+                all_attentions = all_attentions + (layer_outputs[1],)
+
+        if output_hidden_states:
+            encoder_states = encoder_states + (hidden_states,)
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None)
+        return BaseModelOutput(
+            last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions
+        )
+
+
+class MplugOwlVisionModel(PreTrainedModel):
+    main_input_name = "pixel_values"
+    _no_split_modules = ["MplugOwlVisionEncoderLayer"]
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.config = config
+        self.hidden_size = config.hidden_size
+
+        self.embeddings = MplugOwlVisionEmbeddings(config)
+        self.encoder = MplugOwlVisionEncoder(config)
+        self.post_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+
+        self.post_init()
+
+
+    def forward(
+        self,
+        pixel_values: Optional[torch.FloatTensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, BaseModelOutputWithPooling]:
+        r"""
+        Returns:
+
+        """
+        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
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if pixel_values is None:
+            raise ValueError("You have to specify pixel_values")
+
+        hidden_states = self.embeddings(pixel_values)
+
+        encoder_outputs = self.encoder(
+            inputs_embeds=hidden_states,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        last_hidden_state = encoder_outputs[0]
+        last_hidden_state = self.post_layernorm(last_hidden_state)
+
+        pooled_output = last_hidden_state[:, 0, :]
+        pooled_output = self.post_layernorm(pooled_output)
+
+        if not return_dict:
+            return (last_hidden_state, pooled_output) + encoder_outputs[1:]
+
+        return BaseModelOutputWithPooling(
+            last_hidden_state=last_hidden_state,
+            pooler_output=pooled_output,
+            hidden_states=encoder_outputs.hidden_states,
+            attentions=encoder_outputs.attentions,
+        )
+
+    def get_input_embeddings(self):
+        return self.embeddings
+
+
+class MplugOwlVisualAbstractorMLP(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        in_features = config.hidden_size
+        self.act = nn.SiLU()
+
+        self.w1 = nn.Linear(in_features, config.intermediate_size)
+        self.w2 = nn.Linear(config.intermediate_size, in_features)
+        self.w3 = nn.Linear(in_features, config.intermediate_size)
+        self.ffn_ln = nn.LayerNorm(config.intermediate_size, eps=config.layer_norm_eps)
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.act(self.w1(hidden_states)) * self.w3(hidden_states)
+        hidden_states = self.ffn_ln(hidden_states)
+        hidden_states = self.w2(hidden_states)
+        return hidden_states
+
+
+class MplugOwlVisualAbstractorMultiHeadAttention(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        if config.hidden_size % config.num_attention_heads != 0:
+            raise ValueError(
+                "The hidden size (%d) is not a multiple of the number of attention heads (%d)"
+                % (config.hidden_size, config.num_attention_heads)
+            )
+
+        self.num_attention_heads = config.num_attention_heads
+        self.attention_head_size = int(config.hidden_size / config.num_attention_heads)
+        self.all_head_size = self.num_attention_heads * self.attention_head_size
+
+        self.query = nn.Linear(config.hidden_size, self.all_head_size)
+        self.key = nn.Linear(config.encoder_hidden_size, self.all_head_size)
+        self.value = nn.Linear(config.encoder_hidden_size, self.all_head_size)
+
+        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
+        self.save_attention = False
+        
+#         self.q_pos_embed = nn.Parameter(
+#             torch.from_numpy(get_1d_sincos_pos_embed_from_grid(config.hidden_size, np.arange(config.num_learnable_queries, dtype=np.float32))).float()
+#         ).requires_grad_(False)
+#         grids = config.grid_size
+#         self.k_pos_embed = nn.Parameter(
+#             torch.from_numpy(get_2d_sincos_pos_embed(config.hidden_size, grids, cls_token=True)).float()
+#         ).requires_grad_(False)
+        grids = config.grid_size
+        self.register_buffer(
+            'q_pos_embed', 
+            torch.from_numpy(get_1d_sincos_pos_embed_from_grid(config.hidden_size, np.arange(config.num_learnable_queries, dtype=np.float32))).float()
+        )
+        self.register_buffer(
+            'k_pos_embed', 
+            torch.from_numpy(get_2d_sincos_pos_embed(config.hidden_size, grids, cls_token=True)).float()
+        )
+        
+
+    def save_attn_gradients(self, attn_gradients):
+        self.attn_gradients = attn_gradients
+
+    def get_attn_gradients(self):
+        return self.attn_gradients
+
+    def save_attention_map(self, attention_map):
+        self.attention_map = attention_map
+
+    def get_attention_map(self):
+        return self.attention_map
+
+    def transpose_for_scores(self, x):
+        new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.attention_head_size)
+        x = x.view(*new_x_shape)
+        return x.permute(0, 2, 1, 3)
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        head_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        past_key_value=None,
+        output_attentions=False,
+    ):
+        # If this is instantiated as a cross-attention module, the keys
+        # and values come from an encoder; the attention mask needs to be
+        # such that the encoder's padding tokens are not attended to.
+        
+        qk_pos_embed = torch.cat([self.q_pos_embed, self.k_pos_embed], dim = 0).unsqueeze(0).to(dtype=hidden_states.dtype)
+        
+        key_layer = self.transpose_for_scores(self.key(encoder_hidden_states + qk_pos_embed))
+        value_layer = self.transpose_for_scores(self.value(encoder_hidden_states))
+        attention_mask = encoder_attention_mask
+
+        mixed_query_layer = self.query(hidden_states + self.q_pos_embed.unsqueeze(0).to(dtype=hidden_states.dtype))
+
+        query_layer = self.transpose_for_scores(mixed_query_layer)
+
+        past_key_value = (key_layer, value_layer)
+
+        # Take the dot product between "query" and "key" to get the raw attention scores.
+        attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2))
+
+        attention_scores = attention_scores / math.sqrt(self.attention_head_size)
+
+        if attention_mask is not None:
+            # Apply the attention mask is (precomputed for all layers in BertModel forward() function)
+            attention_scores = attention_scores + attention_mask
+
+        # Normalize the attention scores to probabilities.
+        attention_probs = nn.Softmax(dim=-1)(attention_scores)
+
+        if self.save_attention:
+            self.save_attention_map(attention_probs)
+            attention_probs.register_hook(self.save_attn_gradients)
+
+        # This is actually dropping out entire tokens to attend to, which might
+        # seem a bit unusual, but is taken from the original Transformer paper.
+        attention_probs_dropped = self.dropout(attention_probs)
+
+        # Mask heads if we want to
+        if head_mask is not None:
+            attention_probs_dropped = attention_probs_dropped * head_mask
+
+        context_layer = torch.matmul(attention_probs_dropped, value_layer)
+
+        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
+        new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,)
+        context_layer = context_layer.view(*new_context_layer_shape)
+
+        outputs = (context_layer, attention_probs) if output_attentions else (context_layer,)
+
+        outputs = outputs + (past_key_value,)
+        return outputs
+
+
+class MplugOwlVisualAbstractorCrossOutput(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        dim = config.hidden_size
+        self.out_proj = nn.Linear(dim, dim, bias=True)
+        self.norm2 = nn.LayerNorm(dim)
+        self.mlp = MplugOwlVisualAbstractorMLP(config)
+
+    def forward(self, hidden_states: torch.Tensor, input_tensor: torch.Tensor) -> torch.Tensor:
+        input_tensor = input_tensor + self.out_proj(hidden_states)
+        input_tensor = input_tensor + self.mlp(self.norm2(input_tensor))
+        return input_tensor
+
+
+class MplugOwlVisualAbstractorAttention(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.attention = MplugOwlVisualAbstractorMultiHeadAttention(config)
+        self.output = MplugOwlVisualAbstractorCrossOutput(config)
+        self.pruned_heads = set()
+        self.norm1 = nn.LayerNorm(config.hidden_size)
+        self.normk = nn.LayerNorm(config.hidden_size)
+
+    def prune_heads(self, heads):
+        if len(heads) == 0:
+            return
+        heads, index = find_pruneable_heads_and_indices(
+            heads, self.attention.num_attention_heads, self.attention.attention_head_size, self.pruned_heads
+        )
+
+        # Prune linear layers
+        self.attention.query = prune_linear_layer(self.attention.query, index)
+        self.attention.key = prune_linear_layer(self.attention.key, index)
+        self.attention.value = prune_linear_layer(self.attention.value, index)
+        self.output.dense = prune_linear_layer(self.output.out_proj, index, dim=1)
+
+        # Update hyper params and store pruned heads
+        self.attention.num_attention_heads = self.attention.num_attention_heads - len(heads)
+        self.attention.all_head_size = self.attention.attention_head_size * self.attention.num_attention_heads
+        self.pruned_heads = self.pruned_heads.union(heads)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.FloatTensor] = None,
+        encoder_attention_mask: Optional[torch.FloatTensor] = None,
+        past_key_value: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.Tensor]:
+        # HACK we apply norm on q and k
+        hidden_states = self.norm1(hidden_states)
+        encoder_hidden_states = self.normk(encoder_hidden_states)
+        encoder_hidden_states = torch.cat([hidden_states, encoder_hidden_states], dim=1)
+        encoder_attention_mask = torch.cat([attention_mask, encoder_attention_mask], dim=-1)
+        self_outputs = self.attention(
+            hidden_states,
+            attention_mask,
+            head_mask,
+            encoder_hidden_states,
+            encoder_attention_mask,
+            past_key_value,
+            output_attentions,
+        )
+        attention_output = self.output(self_outputs[0], hidden_states)
+        # add attentions if we output them
+        outputs = (attention_output,) + self_outputs[1:]
+        return outputs
+
+
+class MplugOwlVisualAbstractorLayer(nn.Module):
+    def __init__(self, config, layer_idx):
+        super().__init__()
+        self.chunk_size_feed_forward = config.chunk_size_feed_forward
+        self.seq_len_dim = 1
+
+        self.layer_idx = layer_idx
+
+        self.crossattention = MplugOwlVisualAbstractorAttention(config)
+        self.has_cross_attention = True
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        head_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        output_attentions=False,
+    ):
+        if encoder_hidden_states is None:
+            raise ValueError("encoder_hidden_states must be given for cross-attention layers")
+        cross_attention_outputs = self.crossattention(
+            hidden_states,
+            attention_mask,
+            head_mask,
+            encoder_hidden_states,
+            encoder_attention_mask,
+            output_attentions=output_attentions,
+        )
+        query_attention_output = cross_attention_outputs[0]
+
+        outputs = (query_attention_output,)
+        return outputs
+
+
+class MplugOwlVisualAbstractorEncoder(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.layers = nn.ModuleList(
+            [MplugOwlVisualAbstractorLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
+        )
+        self.gradient_checkpointing = True
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        head_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        past_key_values=None,
+        output_attentions=False,
+        output_hidden_states=False,
+        return_dict=True,
+    ):
+        all_hidden_states = () if output_hidden_states else None
+
+        for i in range(self.config.num_hidden_layers):
+            layer_module = self.layers[i]
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)
+
+            layer_head_mask = head_mask[i] if head_mask is not None else None
+            past_key_value = past_key_values[i] if past_key_values is not None else None
+
+            if getattr(self.config, "gradient_checkpointing", False) and self.training:
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        return module(*inputs, past_key_value, output_attentions)
+
+                    return custom_forward
+
+                layer_outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(layer_module),
+                    hidden_states,
+                    attention_mask,
+                    layer_head_mask,
+                    encoder_hidden_states,
+                    encoder_attention_mask,
+                )
+            else:
+                layer_outputs = layer_module(
+                    hidden_states,
+                    attention_mask,
+                    layer_head_mask,
+                    encoder_hidden_states,
+                    encoder_attention_mask,
+                    output_attentions,
+                )
+
+            hidden_states = layer_outputs[0]
+
+        return BaseModelOutput(
+            last_hidden_state=hidden_states,
+        )
+
+
+class MplugOwlVisualAbstractorModel(PreTrainedModel):
+    _no_split_modules = ["MplugOwlVisualAbstractorLayer"]
+    def __init__(self, config, language_hidden_size):
+        super().__init__(config)
+        self.config = config
+
+        self.encoder = MplugOwlVisualAbstractorEncoder(config)
+        self.visual_fc = torch.nn.Linear(config.hidden_size, language_hidden_size)
+        self.query_embeds = torch.nn.Parameter(torch.randn(1, config.num_learnable_queries, config.hidden_size))
+        self.vit_eos = torch.nn.Parameter(torch.randn(1, 1, language_hidden_size))
+
+        self.post_init()
+
+    def _prune_heads(self, heads_to_prune):
+        """
+        Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base
+        class PreTrainedModel
+        """
+        for layer, heads in heads_to_prune.items():
+            self.encoder.layer[layer].attention.prune_heads(heads)
+
+    def get_extended_attention_mask(
+        self,
+        attention_mask: torch.Tensor,
+        input_shape: Tuple[int],
+        device: torch.device,
+    ) -> torch.Tensor:
+        """
+        Makes broadcastable attention and causal masks so that future and masked tokens are ignored.
+
+        Arguments:
+            attention_mask (`torch.Tensor`):
+                Mask with ones indicating tokens to attend to, zeros for tokens to ignore.
+            input_shape (`Tuple[int]`):
+                The shape of the input to the model.
+            device: (`torch.device`):
+                The device of the input to the model.
+
+        Returns:
+            `torch.Tensor` The extended attention mask, with a the same dtype as `attention_mask.dtype`.
+        """
+        # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length]
+        # ourselves in which case we just need to make it broadcastable to all heads.
+        if attention_mask.dim() == 3:
+            extended_attention_mask = attention_mask[:, None, :, :]
+        elif attention_mask.dim() == 2:
+            # Provided a padding mask of dimensions [batch_size, seq_length]
+            # - the model is an encoder, so make the mask broadcastable to [batch_size, num_heads, seq_length, seq_length]
+            extended_attention_mask = attention_mask[:, None, None, :]
+        else:
+            raise ValueError(
+                "Wrong shape for input_ids (shape {}) or attention_mask (shape {})".format(
+                    input_shape, attention_mask.shape
+                )
+            )
+
+        # Since attention_mask is 1.0 for positions we want to attend and 0.0 for
+        # masked positions, this operation will create a tensor which is 0.0 for
+        # positions we want to attend and -10000.0 for masked positions.
+        # Since we are adding it to the raw scores before the softmax, this is
+        # effectively the same as removing these entirely.
+        extended_attention_mask = extended_attention_mask.to(dtype=self.dtype)  # fp16 compatibility
+        extended_attention_mask = (1.0 - extended_attention_mask) * -10000.0
+        return extended_attention_mask
+
+    def forward(
+        self,
+        attention_mask=None,
+        head_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        past_key_values=None,
+        output_attentions=None,
+        output_hidden_states=None,
+        return_dict=None,
+    ):
+        r"""
+        encoder_hidden_states  (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, `optional`):
+            Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if
+            the model is configured as a decoder.
+        encoder_attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, `optional`):
+            Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in
+            the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+        past_key_values (`tuple(tuple(torch.FloatTensor))` of length `config.n_layers` with each tuple having 4 tensors of:
+            shape `(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`): Contains precomputed key and
+            value hidden states of the attention blocks. Can be used to speed up decoding. If `past_key_values` are
+            used, the user can optionally input only the last `decoder_input_ids` (those that don't have their past key
+            value states given to this model) of shape `(batch_size, 1)` instead of all `decoder_input_ids` of shape
+            `(batch_size, sequence_length)`.
+        """
+        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
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+        
+        query_embeds = self.query_embeds.repeat(encoder_hidden_states.shape[0], 1, 1)
+        embedding_output = query_embeds
+        input_shape = embedding_output.size()[:-1]
+        batch_size, seq_length = input_shape
+        device = embedding_output.device
+
+        # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length]
+        # ourselves in which case we just need to make it broadcastable to all heads.
+        if attention_mask is None:
+            attention_mask = torch.ones(
+                (query_embeds.shape[0], query_embeds.shape[1]), dtype=torch.long, device=query_embeds.device
+            )
+        extended_attention_mask = self.get_extended_attention_mask(attention_mask, input_shape, device)
+
+        # If a 2D or 3D attention mask is provided for the cross-attention
+        # we need to make broadcastable to [batch_size, num_heads, seq_length, seq_length]
+        if encoder_hidden_states is not None:
+            if type(encoder_hidden_states) == list:
+                encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states[0].size()
+            else:
+                (
+                    encoder_batch_size,
+                    encoder_sequence_length,
+                    _,
+                ) = encoder_hidden_states.size()
+            encoder_hidden_shape = (encoder_batch_size, encoder_sequence_length)
+
+            if type(encoder_attention_mask) == list:
+                encoder_extended_attention_mask = [self.invert_attention_mask(mask) for mask in encoder_attention_mask]
+            elif encoder_attention_mask is None:
+                encoder_attention_mask = torch.ones(encoder_hidden_shape, device=device)
+                encoder_extended_attention_mask = self.invert_attention_mask(encoder_attention_mask)
+            else:
+                encoder_extended_attention_mask = self.invert_attention_mask(encoder_attention_mask)
+        else:
+            encoder_extended_attention_mask = None
+
+        # Prepare head mask if needed
+        # 1.0 in head_mask indicate we keep the head
+        # attention_probs has shape bsz x n_heads x N x N
+        # input head_mask has shape [num_heads] or [num_hidden_layers x num_heads]
+        # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
+        head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)
+
+        encoder_outputs = self.encoder(
+            embedding_output,
+            attention_mask=extended_attention_mask,
+            head_mask=head_mask,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_extended_attention_mask,
+            past_key_values=past_key_values,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        sequence_output = encoder_outputs[0]
+        pooled_output = sequence_output[:, 0, :]
+
+        sequence_output = self.visual_fc(sequence_output)
+        sequence_output = torch.cat([sequence_output, self.vit_eos.repeat(sequence_output.shape[0], 1, 1)], dim=1)
+
+        return BaseModelOutputWithPooling(
+            last_hidden_state=sequence_output,
+            pooler_output=pooled_output,
+            hidden_states=encoder_outputs.hidden_states,
+        )
+    
+