team-10/venv/Lib/site-packages/transformers/models/aria/modular_aria.py
2025-08-02 02:00:33 +02:00

1665 lines
72 KiB
Python

# coding=utf-8
# Copyright 2024 The Rhymes-AI Teams Authors and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from collections.abc import Iterable
from typing import Optional, Union
import numpy as np
from ...activations import ACT2FN
from ...cache_utils import Cache
from ...configuration_utils import PretrainedConfig
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_patch_output_size, select_best_resolution
from ...image_transforms import PaddingMode, convert_to_rgb, pad, resize, to_channel_dimension_format
from ...image_utils import (
ChannelDimension,
ImageInput,
PILImageResampling,
get_image_size,
infer_channel_dimension_format,
is_scaled_image,
make_flat_list_of_images,
to_numpy_array,
valid_images,
validate_preprocess_arguments,
)
from ...modeling_flash_attention_utils import FlashAttentionKwargs
from ...modeling_utils import PreTrainedModel
from ...processing_utils import MultiModalData, ProcessingKwargs, ProcessorMixin, Unpack
from ...tokenization_utils import PreTokenizedInput, TextInput
from ...utils import TensorType, TransformersKwargs, auto_docstring, can_return_tuple, logging
from ...utils.import_utils import is_torch_available
from ..auto import CONFIG_MAPPING, AutoConfig, AutoTokenizer
from ..llama.configuration_llama import LlamaConfig
from ..llama.modeling_llama import (
LlamaAttention,
LlamaDecoderLayer,
LlamaForCausalLM,
LlamaMLP,
LlamaModel,
LlamaPreTrainedModel,
LlamaRMSNorm,
)
from ..llava.modeling_llava import (
LlavaCausalLMOutputWithPast,
LlavaForConditionalGeneration,
LlavaModel,
LlavaModelOutputWithPast,
)
from ..llava_next.image_processing_llava_next import divide_to_patches
logger = logging.get_logger(__name__)
if is_torch_available():
import torch
from torch import nn
def sequential_experts_gemm(token_states, expert_weights, tokens_per_expert):
"""
Compute the matrix multiplication (GEMM) for each expert sequentially. This approach is computationally inefficient, especially when dealing with a large number of experts.
Args:
token_states (torch.Tensor): Input tensor of shape (num_tokens, in_features).
expert_weights (torch.Tensor): Weight tensor of shape (num_experts, in_features, out_features).
tokens_per_expert (torch.Tensor): Number of tokens assigned to each expert.
Returns:
torch.Tensor: Output tensor of shape (num_tokens, out_features).
"""
num_tokens = token_states.shape[0]
out_features = expert_weights.shape[-1]
output = torch.zeros(num_tokens, out_features, dtype=token_states.dtype, device=token_states.device)
cumsum_num_tokens = torch.cumsum(tokens_per_expert, dim=0)
# Insert zero at the beginning for offset index's convenience
zero_tensor = torch.zeros(1, dtype=torch.long, device=cumsum_num_tokens.device)
cumsum_num_tokens = torch.cat((zero_tensor, cumsum_num_tokens))
for expert_num in range(expert_weights.shape[0]):
start = cumsum_num_tokens[expert_num]
end = cumsum_num_tokens[expert_num + 1]
tokens = token_states[start:end]
out = torch.matmul(tokens, expert_weights[expert_num])
output[start:end] = out
return output
class AriaTextConfig(LlamaConfig):
r"""
This class handles the configuration for the text component of the Aria model.
Instantiating a configuration with the defaults will yield a similar configuration to that of the model of the Aria
[rhymes-ai/Aria](https://huggingface.co/rhymes-ai/Aria) architecture.
This class extends the LlamaConfig to include additional parameters specific to the Mixture of Experts (MoE) architecture.
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 4096):
The size 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, check out [this
paper](https://huggingface.co/papers/2305.13245). 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*, defaults to 2):
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/main/perf_train_gpu_many#tensor-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. NOTE: if you apply new rope type
and you expect the model to work on longer `max_position_embeddings`, we recommend you to update this value
accordingly.
Expected contents:
`rope_type` (`str`):
The sub-variant of RoPE to use. Can be one of ['default', 'linear', 'dynamic', 'yarn', 'longrope',
'llama3'], with 'default' being the original RoPE implementation.
`factor` (`float`, *optional*):
Used with all rope types except 'default'. The scaling factor to apply to the RoPE embeddings. In
most scaling types, a `factor` of x will enable the model to handle sequences of length x *
original maximum pre-trained length.
`original_max_position_embeddings` (`int`, *optional*):
Used with 'dynamic', 'longrope' and 'llama3'. The original max position embeddings used during
pretraining.
`attention_factor` (`float`, *optional*):
Used with 'yarn' and 'longrope'. The scaling factor to be applied on the attention
computation. If unspecified, it defaults to value recommended by the implementation, using the
`factor` field to infer the suggested value.
`beta_fast` (`float`, *optional*):
Only used with 'yarn'. Parameter to set the boundary for extrapolation (only) in the linear
ramp function. If unspecified, it defaults to 32.
`beta_slow` (`float`, *optional*):
Only used with 'yarn'. Parameter to set the boundary for interpolation (only) in the linear
ramp function. If unspecified, it defaults to 1.
`short_factor` (`list[float]`, *optional*):
Only used with 'longrope'. The scaling factor to be applied to short contexts (<
`original_max_position_embeddings`). Must be a list of numbers with the same length as the hidden
size divided by the number of attention heads divided by 2
`long_factor` (`list[float]`, *optional*):
Only used with 'longrope'. The scaling factor to be applied to long contexts (<
`original_max_position_embeddings`). Must be a list of numbers with the same length as the hidden
size divided by the number of attention heads divided by 2
`low_freq_factor` (`float`, *optional*):
Only used with 'llama3'. Scaling factor applied to low frequency components of the RoPE
`high_freq_factor` (`float`, *optional*):
Only used with 'llama3'. Scaling factor applied to high frequency components of the RoPE
attention_bias (`bool`, *optional*, defaults to `False`):
Whether to use a bias in the query, key, value and output projection layers during self-attention.
attention_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for the attention probabilities.
mlp_bias (`bool`, *optional*, defaults to `False`):
Whether to use a bias in up_proj, down_proj and gate_proj layers in the MLP layers.
head_dim (`int`, *optional*):
The attention head dimension. If None, it will default to hidden_size // num_heads
moe_num_experts (`int`, *optional*, defaults to 8):
The number of experts in the MoE layer.
moe_topk (`int`, *optional*, defaults to 2):
The number of top experts to route to for each token.
moe_num_shared_experts (`int`, *optional*, defaults to 2):
The number of shared experts.
"""
model_type = "aria_text"
base_config_key = "text_config"
def __init__(
self,
intermediate_size: int = 4096,
moe_num_experts: int = 8,
moe_topk: int = 2,
moe_num_shared_experts: int = 2,
pad_token_id=2,
**super_kwargs,
):
super().__init__(pad_token_id=pad_token_id, **super_kwargs)
self.intermediate_size = intermediate_size
self.moe_num_experts = moe_num_experts
self.moe_topk = moe_topk
self.moe_num_shared_experts = moe_num_shared_experts
class AriaConfig(PretrainedConfig):
r"""
This class handles the configuration for both vision and text components of the Aria model,
as well as additional parameters for image token handling and projector mapping.
Instantiating a configuration with the defaults will yield a similar configuration to that of the model of the Aria
[rhymes-ai/Aria](https://huggingface.co/rhymes-ai/Aria) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
vision_config (`AriaVisionConfig` or `dict`, *optional*):
Configuration for the vision component.
vision_feature_layer (`int`, *optional*, defaults to -1):
The index of the layer to select the vision feature.
text_config (`AriaTextConfig` or `dict`, *optional*):
Configuration for the text component.
projector_patch_to_query_dict (`dict`, *optional*):
Mapping of patch sizes to query dimensions.
image_token_index (`int`, *optional*, defaults to 9):
Index used to represent image tokens.
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated normal initializer for initializing all weight matrices.
Attributes:
model_type (`str`):
Type of the model, set to `"aria"`.
image_token_index (`int`):
Index used to represent image tokens.
projector_patch_to_query_dict (`dict`):
Mapping of patch sizes to query dimensions.
vision_config (`AriaVisionConfig`):
Configuration for the vision component.
text_config (`AriaTextConfig`):
Configuration for the text component.
"""
model_type = "aria"
attribute_map = {
"image_token_id": "image_token_index",
}
sub_configs = {"text_config": AriaTextConfig, "vision_config": AutoConfig}
def __init__(
self,
vision_config=None,
vision_feature_layer: int = -1,
text_config: AriaTextConfig = None,
projector_patch_to_query_dict: Optional[dict] = None,
image_token_index: int = 9,
initializer_range: float = 0.02,
**kwargs,
):
self.image_token_index = image_token_index
# Convert the keys and values of projector_patch_to_query_dict to integers
# This ensures consistency even if they were provided as strings
if projector_patch_to_query_dict is None:
projector_patch_to_query_dict = {
1225: 128,
4900: 256,
}
self.projector_patch_to_query_dict = {int(k): int(v) for k, v in projector_patch_to_query_dict.items()}
self.max_value_projector_patch_to_query_dict = max(self.projector_patch_to_query_dict.values())
self.vision_feature_layer = vision_feature_layer
if isinstance(vision_config, dict):
vision_config["model_type"] = "idefics3_vision"
vision_config = CONFIG_MAPPING[vision_config["model_type"]](**vision_config)
elif vision_config is None:
vision_config = CONFIG_MAPPING["idefics3_vision"]()
self.vision_config = vision_config
self.initializer_range = initializer_range
if isinstance(text_config, dict) and "model_type" in text_config:
text_config = AriaTextConfig(**text_config)
elif text_config is None:
text_config = AriaTextConfig()
self.text_config = text_config
super().__init__(**kwargs)
class AriaTextRMSNorm(LlamaRMSNorm):
pass
class AriaProjectorMLP(nn.Module):
"""
Feed-Forward Network module for the Aria Projector.
Args:
in_features (`int`):
Input embedding dimension.
hidden_features (`int`):
Hidden dimension of the feed-forward network.
output_dim (`int`):
Output dimension.
"""
def __init__(self, in_features, hidden_features, output_dim):
super().__init__()
self.linear_in = nn.Linear(in_features, hidden_features, bias=False)
self.linear_out = nn.Linear(hidden_features, output_dim, bias=False)
self.act = ACT2FN["gelu_new"]
def forward(self, hidden_states):
hidden_states = self.act(self.linear_in(hidden_states))
hidden_states = self.linear_out(hidden_states)
return hidden_states
class AriaCrossAttention(nn.Module):
"""
Aria Cross-Attention module.
Args:
config (`AriaConfig`):
The configuration to use.
"""
def __init__(self, config: AriaConfig, dropout_rate: float = 0):
super().__init__()
hidden_size = config.vision_config.hidden_size
num_heads = config.vision_config.num_attention_heads
self.num_heads = num_heads
self.q_proj = nn.Linear(hidden_size, hidden_size, bias=False)
self.k_proj = nn.Linear(hidden_size, hidden_size, bias=False)
self.v_proj = nn.Linear(hidden_size, hidden_size, bias=False)
# Original code here: https://github.com/rhymes-ai/Aria/blob/719ff4e52b727443cba3793b0e27fe64e0244fe1/aria/model/projector.py#L48
self.multihead_attn = nn.MultiheadAttention(hidden_size, num_heads, batch_first=True)
self.linear = nn.Linear(hidden_size, hidden_size)
self.dropout = nn.Dropout(dropout_rate)
self.layer_norm = nn.LayerNorm(hidden_size)
self.layer_norm_kv = nn.LayerNorm(hidden_size)
def forward(self, key_value_states, hidden_states, attn_mask=None):
"""
Forward pass of the AriaCrossAttention module.
Args:
key_value_states (`torch.Tensor`):
Input tensor for key and value.
hidden_states (`torch.Tensor`):
Input tensor for query.
attn_mask (`torch.Tensor`, *optional*, defaults to None):
Attention mask.
Returns:
torch.Tensor:
Output tensor after cross-attention.
"""
query = self.q_proj(self.layer_norm(hidden_states))
key_value_states = self.layer_norm_kv(key_value_states)
key = self.k_proj(key_value_states)
value = self.v_proj(key_value_states)
attn_output, _ = self.multihead_attn(query, key, value, attn_mask=attn_mask)
attn_output = self.dropout(self.linear(attn_output))
return attn_output
class AriaProjector(nn.Module):
"""
Aria Projector module.
This module projects vision features into the language model's embedding space, enabling interaction between vision and language components.
Args:
config (`AriaConfig`):
Configuration object for the model.
"""
def __init__(
self,
config: AriaConfig,
):
super().__init__()
self.patch_to_query_dict = config.projector_patch_to_query_dict
self.in_features = config.vision_config.hidden_size
self.num_heads = config.vision_config.num_attention_heads
self.kv_dim = config.vision_config.hidden_size
self.hidden_features = config.text_config.hidden_size
self.output_dim = config.text_config.hidden_size
self.query = nn.Parameter(torch.zeros(config.max_value_projector_patch_to_query_dict, self.in_features))
self.cross_attn = AriaCrossAttention(config)
self.layer_norm = nn.LayerNorm(self.in_features)
self.feed_forward = AriaProjectorMLP(self.in_features, self.hidden_features, self.output_dim)
def forward(self, key_value_states: torch.Tensor, attn_mask: Optional[torch.Tensor] = None):
"""
Forward pass of the Projector module.
Args:
key_value_states (`torch.Tensor`):
Input tensor of shape (batch_size, num_patches, kv_dim).
attn_mask (`torch.Tensor`, *optional*, default is None):
Attention mask.
Returns:
`torch.Tensor`: Output tensor of shape (batch_size, query_number, output_dim).
"""
batch_size, num_patches = key_value_states.shape[0], key_value_states.shape[1]
if num_patches not in self.patch_to_query_dict.keys():
raise KeyError(
f"Number of patches {num_patches} not found in patch_to_query_dict amongst possible values {self.patch_to_query_dict.keys()}."
)
query_num = self.patch_to_query_dict[num_patches]
queries = self.query[:query_num].unsqueeze(0).repeat(batch_size, 1, 1)
if attn_mask is not None:
attn_mask = attn_mask.repeat_interleave(self.num_heads, 0)
attn_mask = attn_mask.unsqueeze(1).expand(-1, queries.size(1), -1)
attention_out = self.cross_attn(key_value_states, queries, attn_mask=attn_mask)
out = self.feed_forward(self.layer_norm(attention_out))
return out
class AriaImageProcessor(BaseImageProcessor):
"""
A vision processor for the Aria model that handles image preprocessing.
Initialize the AriaImageProcessor.
Args:
image_mean (`list`, *optional*, defaults to [0.5, 0.5, 0.5]):
Mean values for normalization.
image_std (`list`, *optional*, defaults to [0.5, 0.5, 0.5]):
Standard deviation values for normalization.
max_image_size (`int`, *optional*, defaults to 980):
Maximum image size.
min_image_size (`int`, *optional*, defaults to 336):
Minimum image size.
split_resolutions (`list`, *optional*, defaults to a list of optimal,resolutions as tuples):
The optimal resolutions for splitting the image.
split_image (`bool`, *optional*, defaults to `False`):
Whether to split the image.
do_convert_rgb (`bool`, *optional*, defaults to `True`):
Whether to convert the image to RGB.
do_rescale (`bool`, *optional*, defaults to `True`):
Whether to rescale the image by the specified scale `rescale_factor`. Can be overridden by `do_rescale` in
the `preprocess` method.
rescale_factor (`int` or `float`, *optional*, defaults to `1/255`):
Scale factor to use if rescaling the image. Can be overridden by `rescale_factor` in the `preprocess`
method.
do_normalize (`bool`, *optional*, defaults to `True`):
Whether to normalize the image.
resample (PILImageResampling, *optional*, defaults to `BICUBIC`):
The resampling filter to use if resizing the image.
"""
model_input_names = ["pixel_values", "pixel_mask", "num_crops"]
def __init__(
self,
image_mean: Optional[list[float]] = None,
image_std: Optional[list[float]] = None,
max_image_size: int = 980,
min_image_size: int = 336,
split_resolutions: Optional[list[tuple[int, int]]] = None,
split_image: Optional[bool] = False,
do_convert_rgb: Optional[bool] = True,
do_rescale: bool = True,
rescale_factor: Union[int, float] = 1 / 255,
do_normalize: Optional[bool] = True,
resample: PILImageResampling = PILImageResampling.BICUBIC,
**kwargs,
):
super().__init__(**kwargs)
if image_mean is None:
image_mean = [0.5, 0.5, 0.5]
if image_std is None:
image_std = [0.5, 0.5, 0.5]
self.max_image_size = max_image_size
self.min_image_size = min_image_size
self.image_mean = image_mean
self.image_std = image_std
self.split_image = split_image
if split_resolutions is None:
split_resolutions = [(1, 2), (1, 3), (1, 4), (1, 5), (1, 6), (1, 7), (1, 8), (2, 4), (2, 3), (2, 2), (2, 1), (3, 1), (3, 2), (4, 1), (4, 2), (5, 1), (6, 1), (7, 1), (8, 1)] # fmt: skip
split_resolutions = [(el[0] * 490, el[1] * 490) for el in split_resolutions]
self.split_resolutions = split_resolutions
self.do_convert_rgb = do_convert_rgb
self.do_rescale = do_rescale
self.rescale_factor = rescale_factor
self.do_normalize = do_normalize
self.resample = resample
def preprocess(
self,
images: Union[ImageInput, list[ImageInput]],
image_mean: Optional[Union[float, list[float]]] = None,
image_std: Optional[Union[float, list[float]]] = None,
max_image_size: Optional[int] = None,
min_image_size: Optional[int] = None,
split_image: Optional[bool] = None,
do_convert_rgb: Optional[bool] = None,
do_rescale: Optional[bool] = None,
rescale_factor: Optional[float] = None,
do_normalize: Optional[bool] = None,
resample: PILImageResampling = None,
return_tensors: Optional[Union[str, TensorType]] = "pt",
data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
input_data_format: Optional[Union[str, ChannelDimension]] = None,
):
"""
Process a list of images.
Args:
images (ImageInput or list of ImageInput):
The input image or a list of images.
image_mean (`list`, *optional*, defaults to [0.5, 0.5, 0.5]):
Mean values for normalization.
image_std (`list`, *optional*, defaults to [0.5, 0.5, 0.5]):
Standard deviation values for normalization.
max_image_size (`int`, *optional*, defaults to `self.max_image_size` (980)):
Maximum image size.
min_image_size (`int`, *optional*, defaults to `self.min_image_size` (336)):
Minimum image size.
split_image (`bool`, *optional*, defaults to `self.split_image` (False)):
Whether to split the image.
do_convert_rgb (`bool`, *optional*, defaults to `self.do_convert_rgb` (True)):
Whether to convert the image to RGB.
do_rescale (`bool`, *optional*, defaults to `self.do_rescale`):
Whether to rescale the image.
rescale_factor (`float`, *optional*, defaults to `self.rescale_factor`):
Rescale factor to rescale the image by if `do_rescale` is set to `True`.
do_normalize (`bool`, *optional*, defaults to `self.do_normalize` (True)):
Whether to normalize the image.
resample (PILImageResampling, *optional*, defaults to `self.resample` (BICUBIC)):
The resampling filter to use if resizing the image.
return_tensors (`str` or `TensorType`, *optional*, defaults to "pt"):
The type of tensor to return.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format for the output image. Can be one of:
- `"channels_first"` or `ChannelDimension.FIRST`:
image in (num_channels, height, width) format.
- `"channels_last"` or `ChannelDimension.LAST`:
image in (height, width, num_channels) format.
If unset, will use same as the input image.
input_data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format for the input image. Can be one of:
- `"channels_first"` or `ChannelDimension.FIRST`:
image in (num_channels, height, width) format.
- `"channels_last"` or `ChannelDimension.LAST`:
image in (height, width, num_channels) format.
If unset, will use the inferred format of the input image.
Returns:
BatchFeature:
A BatchFeature object containing:
- 'pixel_values':
Tensor of processed image pixel values.
- 'pixel_mask':
Boolean pixel mask. This mask is a 2D tensor of shape (max_image_size, max_image_size) where:
- True (1) values indicate pixels that belong to the original resized image.
- False (0) values indicate pixels that are part of the padding.
The mask helps distinguish between actual image content and padded areas in subsequent processing steps.
- 'num_crops':
The maximum number of crops across all images.
"""
image_mean = image_mean if image_mean is not None else self.image_mean
image_std = image_std if image_std is not None else self.image_std
max_image_size = max_image_size if max_image_size is not None else self.max_image_size
min_image_size = min_image_size if min_image_size is not None else self.min_image_size
split_image = split_image if split_image is not None else self.split_image
do_convert_rgb = do_convert_rgb if do_convert_rgb is not None else self.do_convert_rgb
do_rescale = do_rescale if do_rescale is not None else self.do_rescale
rescale_factor = rescale_factor if rescale_factor is not None else self.rescale_factor
do_normalize = do_normalize if do_normalize is not None else self.do_normalize
resample = resample if resample is not None else self.resample
if max_image_size not in [490, 980]:
raise ValueError("max_image_size must be either 490 or 980")
images = make_flat_list_of_images(images)
if not valid_images(images):
raise ValueError(
"Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
"torch.Tensor, tf.Tensor or jax.ndarray."
)
validate_preprocess_arguments(
do_normalize=do_normalize,
image_mean=image_mean,
image_std=image_std,
resample=resample,
do_rescale=do_rescale,
rescale_factor=rescale_factor,
)
if do_convert_rgb:
images = [convert_to_rgb(image) for image in images]
# All transformations expect numpy arrays.
images = [to_numpy_array(image) for image in images]
if do_rescale and is_scaled_image(images[0]):
logger.warning_once(
"It looks like you are trying to rescale already rescaled images. If the input"
" images have pixel values between 0 and 1, set `do_rescale=False` to avoid rescaling them again."
)
if input_data_format is None:
# We assume that all images have the same channel dimension format.
input_data_format = infer_channel_dimension_format(images[0])
pixel_values = []
pixel_masks = []
num_crops = None
for image in images:
if split_image:
crop_images = self.get_image_patches(
image,
self.split_resolutions,
max_image_size,
resample,
data_format=input_data_format,
input_data_format=input_data_format,
)
else:
crop_images = [image]
if num_crops is None or len(crop_images) > num_crops:
num_crops = len(crop_images)
for crop_image in crop_images:
# At this point the scale is the rescaling factor that would bring the image to max_size in its larger dimension
h, w = get_image_size(crop_image)
scale = max_image_size / max(h, w)
if w >= h:
new_size = (max(int(h * scale), min_image_size), max_image_size) # h, w
else:
new_size = (max_image_size, max(int(w * scale), min_image_size)) # h, w
crop_image_resized = resize(
crop_image,
new_size,
resample=resample,
data_format=input_data_format,
input_data_format=input_data_format,
)
padding_bottom, padding_right = max_image_size - new_size[0], max_image_size - new_size[1]
crop_image_padded = pad(
crop_image_resized,
((0, padding_bottom), (0, padding_right)),
data_format=input_data_format,
input_data_format=input_data_format,
)
# Create a pixel mask
pixel_mask = np.zeros((max_image_size, max_image_size), dtype=bool)
pixel_mask[: new_size[0], : new_size[1]] = 1
pixel_masks.append(pixel_mask)
if do_rescale:
crop_image_padded = self.rescale(
image=crop_image_padded, scale=rescale_factor, input_data_format=input_data_format
)
if do_normalize:
crop_image_padded = self.normalize(
crop_image_padded,
self.image_mean,
self.image_std,
data_format=input_data_format,
input_data_format=input_data_format,
)
crop_image_padded = (
to_channel_dimension_format(crop_image_padded, data_format, input_data_format)
if data_format is not None
else crop_image_padded
)
pixel_values.append(crop_image_padded)
return BatchFeature(
data={
"pixel_values": np.stack(pixel_values, axis=0),
"pixel_mask": np.stack(pixel_masks, axis=0),
"num_crops": num_crops,
},
tensor_type=return_tensors,
)
def _resize_for_patching(
self, image: np.array, target_resolution: tuple, resample, input_data_format: ChannelDimension
) -> np.array:
"""
Resizes an image to a target resolution while maintaining aspect ratio.
Args:
image (np.array):
The input image.
target_resolution (tuple):
The target resolution (height, width) of the image.
resample (`PILImageResampling`):
Resampling filter to use if resizing the image.
input_data_format (`ChannelDimension` or `str`):
The channel dimension format of the input image.
Returns:
np.array: The resized and padded image.
"""
new_height, new_width = get_patch_output_size(image, target_resolution, input_data_format)
# Resize the image
resized_image = resize(image, (new_height, new_width), resample=resample, input_data_format=input_data_format)
return resized_image
def _get_padding_size(self, original_resolution: tuple, target_resolution: tuple):
original_height, original_width = original_resolution
target_height, target_width = target_resolution
paste_x, r_x = divmod(target_width - original_width, 2)
paste_y, r_y = divmod(target_height - original_height, 2)
return (paste_y, paste_y + r_y), (paste_x, paste_x + r_x)
def _pad_for_patching(
self, image: np.array, target_resolution: tuple, input_data_format: ChannelDimension
) -> np.array:
"""
Pad an image to a target resolution while maintaining aspect ratio.
"""
new_resolution = get_patch_output_size(image, target_resolution, input_data_format)
padding = self._get_padding_size(new_resolution, target_resolution)
padded_image = self.pad(image, padding=padding)
return padded_image
def pad(
self,
image: np.ndarray,
padding: Union[int, tuple[int, int], Iterable[tuple[int, int]]],
mode: PaddingMode = PaddingMode.CONSTANT,
constant_values: Union[float, Iterable[float]] = 0.0,
data_format: Optional[Union[str, ChannelDimension]] = None,
input_data_format: Optional[Union[str, ChannelDimension]] = None,
) -> np.ndarray:
"""
Pads the `image` with the specified `padding` and `mode`. Padding can be in the (`height`, `width`)
dimension of in the (`num_patches`) dimension. In the second case an iterable if tuples is expected
as input.
Args:
image (`np.ndarray`):
The image to pad.
padding (`int` or `tuple[int, int]` or `Iterable[tuple[int, int]]`):
Padding to apply to the edges of the height, width axes. Can be one of three formats:
- `((before_height, after_height), (before_width, after_width))` unique pad widths for each axis.
- `((before, after),)` yields same before and after pad for height and width.
- `(pad,)` or int is a shortcut for before = after = pad width for all axes.
mode (`PaddingMode`):
The padding mode to use. Can be one of:
- `"constant"`: pads with a constant value.
- `"reflect"`: pads with the reflection of the vector mirrored on the first and last values of the
vector along each axis.
- `"replicate"`: pads with the replication of the last value on the edge of the array along each axis.
- `"symmetric"`: pads with the reflection of the vector mirrored along the edge of the array.
constant_values (`float` or `Iterable[float]`, *optional*):
The value to use for the padding if `mode` is `"constant"`.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format for the output image. Can be one of:
- `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
- `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
If unset, will use same as the input image.
input_data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format for the input image. Can be one of:
- `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
- `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
If unset, will use the inferred format of the input image.
Returns:
`np.ndarray`: The padded image.
"""
# call the general `pad` if padding on `height/width`, otherwise it's the `num_patched` dim
if isinstance(padding, int) or len(padding) != 4:
return pad(image, padding, mode, constant_values, data_format, input_data_format)
if input_data_format is None:
input_data_format = infer_channel_dimension_format(image)
padding_mode_mapping = {
PaddingMode.CONSTANT: "constant",
PaddingMode.REFLECT: "reflect",
PaddingMode.REPLICATE: "edge",
PaddingMode.SYMMETRIC: "symmetric",
}
image = np.pad(image, padding, mode=padding_mode_mapping[mode], constant_values=constant_values)
image = (
to_channel_dimension_format(image, data_format, input_data_format) if data_format is not None else image
)
return image
def get_image_patches(
self,
image: np.array,
grid_pinpoints: list[tuple[int, int]],
patch_size: int,
resample: PILImageResampling,
data_format: ChannelDimension,
input_data_format: ChannelDimension,
) -> list[np.array]:
"""
Process an image with variable resolutions by dividing it into patches.
Args:
image (`np.array`):
The input image to be processed.
grid_pinpoints (list[tuple[int, int]]):
A list of possible resolutions as tuples.
patch_size (`int`):
Size of the patches to divide the image into.
resample (`PILImageResampling`):
Resampling filter to use if resizing the image.
data_format (`ChannelDimension` or `str`):
The channel dimension format for the output image.
input_data_format (`ChannelDimension` or `str`):
The channel dimension format of the input image.
Returns:
`list[np.array]`: A list of NumPy arrays containing the processed image patches.
"""
if not isinstance(grid_pinpoints, list):
raise TypeError("grid_pinpoints must be a list of possible resolutions.")
possible_resolutions = grid_pinpoints
image_size = get_image_size(image, channel_dim=input_data_format)
best_resolution = select_best_resolution(image_size, possible_resolutions)
resized_image = self._resize_for_patching(
image, best_resolution, resample=resample, input_data_format=input_data_format
)
padded_image = self._pad_for_patching(resized_image, best_resolution, input_data_format=input_data_format)
patches = divide_to_patches(padded_image, patch_size=patch_size, input_data_format=input_data_format)
# make sure that all patches are in the input data format
patches = [
to_channel_dimension_format(patch, channel_dim=data_format, input_channel_dim=input_data_format)
for patch in patches
]
return patches
def get_number_of_image_patches(self, height: int, width: int, images_kwargs=None):
"""
A utility that returns number of image patches for a given image size.
Args:
height (`int`):
Height of the input image.
width (`int`):
Width of the input image.
images_kwargs (`dict`, *optional*)
Any kwargs to override defaults of the image processor.
Returns:
`int`: Number of patches per image.
"""
split_image = images_kwargs.get("split_image", None) or self.split_image
max_image_size = images_kwargs.get("max_image_size", None) or self.max_image_size
resized_height, resized_width = select_best_resolution((height, width), self.split_resolutions)
num_patches = 1 if not split_image else resized_height // max_image_size * resized_width // max_image_size
return num_patches
class AriaProcessorKwargs(ProcessingKwargs, total=False):
_defaults = {
"text_kwargs": {
"padding": False,
"return_mm_token_type_ids": False,
},
"images_kwargs": {
"max_image_size": 980,
"split_image": False,
},
"return_tensors": TensorType.PYTORCH,
}
class AriaProcessor(ProcessorMixin):
"""
AriaProcessor is a processor for the Aria model which wraps the Aria image preprocessor and the LLama slow tokenizer.
Args:
image_processor (`AriaImageProcessor`, *optional*):
The AriaImageProcessor to use for image preprocessing.
tokenizer (`PreTrainedTokenizerBase`, *optional*):
An instance of [`PreTrainedTokenizerBase`]. This should correspond with the model's text model. The tokenizer is a required input.
chat_template (`str`, *optional*):
A Jinja template which will be used to convert lists of messages in a chat into a tokenizable string.
size_conversion (`Dict`, *optional*):
A dictionary indicating size conversions for images.
"""
attributes = ["image_processor", "tokenizer"]
image_processor_class = "AriaImageProcessor"
tokenizer_class = "AutoTokenizer"
def __init__(
self,
image_processor=None,
tokenizer: Union[AutoTokenizer, str] = None,
chat_template: Optional[str] = None,
size_conversion: Optional[dict[Union[float, int], int]] = None,
):
if size_conversion is None:
size_conversion = {490: 128, 980: 256}
self.size_conversion = {int(k): v for k, v in size_conversion.items()}
self.image_token = tokenizer.image_token
self.image_token_id = tokenizer.image_token_id
if tokenizer is not None and tokenizer.pad_token is None:
tokenizer.pad_token = tokenizer.unk_token
super().__init__(image_processor, tokenizer, chat_template=chat_template)
def __call__(
self,
text: Union[TextInput, PreTokenizedInput, list[TextInput], list[PreTokenizedInput]],
images: Optional[ImageInput] = None,
audio=None,
videos=None,
**kwargs: Unpack[AriaProcessorKwargs],
) -> BatchFeature:
"""
Main method to prepare for the model one or several sequences(s) and image(s).
Args:
text (`TextInput`, `PreTokenizedInput`, `list[TextInput]`, `list[PreTokenizedInput]`):
The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings
(pretokenized string). If the sequences are provided as list of strings (pretokenized), you must set
`is_split_into_words=True` (to lift the ambiguity with a batch of sequences).
images (`ImageInput`):
The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch
tensor. Both channels-first and channels-last formats are supported.
Returns:
[`BatchFeature`]: A [`BatchFeature`] with the following fields:
- **input_ids** -- List of token ids to be fed to a model. Returned when `text` is not `None`.
- **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when
`return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not
`None`).
- **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`.
- **pixel_mask** -- Pixel mask to be fed to a model. Returned when `images` is not `None`.
"""
output_kwargs = self._merge_kwargs(
AriaProcessorKwargs,
tokenizer_init_kwargs=self.tokenizer.init_kwargs,
**kwargs,
)
if isinstance(text, str):
text = [text]
elif not isinstance(text, list) and not isinstance(text[0], str):
raise TypeError("Invalid input text. Please provide a string, or a list of strings")
if images is not None:
image_inputs = self.image_processor(images, **output_kwargs["images_kwargs"])
# expand the image_token according to the num_crops and tokens per image
tokens_per_image = self.size_conversion[image_inputs.pixel_values.shape[2]]
prompt_strings = []
num_crops = image_inputs.pop("num_crops") * tokens_per_image
for sample in text:
sample = sample.replace(self.tokenizer.image_token, self.tokenizer.image_token * num_crops)
prompt_strings.append(sample)
else:
image_inputs = {}
prompt_strings = text
return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None)
return_mm_token_type_ids = output_kwargs["text_kwargs"].pop("return_mm_token_type_ids", False)
text_inputs = self.tokenizer(prompt_strings, **output_kwargs["text_kwargs"], return_tensors=None)
self._check_special_mm_tokens(prompt_strings, text_inputs, modalities=["image"])
if return_mm_token_type_ids:
array_ids = np.array(text_inputs["input_ids"])
mm_token_type_ids = np.zeros_like(text_inputs["input_ids"])
mm_token_type_ids[array_ids == self.image_token_id] = 1
text_inputs["mm_token_type_ids"] = mm_token_type_ids.tolist()
return BatchFeature(data={**text_inputs, **image_inputs}, tensor_type=return_tensors)
def _get_num_multimodal_tokens(self, image_sizes=None, **kwargs):
"""
Computes the number of placeholder tokens needed for multimodal inputs with the given sizes.
Args:
image_sizes (`list[list[int]]`, *optional*):
The input sizes formatted as (height, width) per each image.
Returns:
`MultiModalData`: A `MultiModalData` object holding number of tokens per each of the provided
input modalities, along with other useful data.
"""
vision_data = {}
if image_sizes is not None:
images_kwargs = AriaProcessorKwargs._defaults.get("images_kwargs", {})
images_kwargs.update(kwargs)
max_size = images_kwargs.get("max_image_size", None) or self.image_processor.max_image_size
num_image_patches = [
self.image_processor.get_number_of_image_patches(*image_size, images_kwargs)
for image_size in image_sizes
]
num_image_tokens = [self.size_conversion[max_size] * num_patches for num_patches in num_image_patches]
vision_data.update({"num_image_tokens": num_image_tokens, "num_image_patches": num_image_patches})
return MultiModalData(**vision_data)
def batch_decode(self, *args, **kwargs):
"""
This method forwards all its arguments to LlamaTokenizerFast's [`~PreTrainedTokenizer.batch_decode`]. Please
refer to the docstring of this method for more information.
"""
return self.tokenizer.batch_decode(*args, **kwargs)
def decode(self, *args, **kwargs):
"""
This method forwards all its arguments to LlamaTokenizerFast's [`~PreTrainedTokenizer.decode`]. Please refer to
the docstring of this method for more information.
"""
return self.tokenizer.decode(*args, **kwargs)
@property
def model_input_names(self):
tokenizer_input_names = self.tokenizer.model_input_names
image_processor_input_names = self.image_processor.model_input_names
# Remove `num_crops`, it is popped and used only when processing. Make a copy of list when remocing
# otherwise `self.image_processor.model_input_names` is also modified
image_processor_input_names = [name for name in image_processor_input_names if name != "num_crops"]
return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))
class AriaSharedExpertsMLP(LlamaMLP):
"""
Shared Expert MLP for shared experts.
Unlike routed experts, shared experts process all tokens without routing.
This class reconfigures the intermediate size in comparison to the LlamaMLP.
Args:
config (`AriaTextConfig`): Configuration object for the Aria language model.
"""
def __init__(self, config: AriaTextConfig):
super().__init__(self)
self.intermediate_size = config.intermediate_size * config.moe_num_shared_experts
class AriaGroupedExpertsGemm(nn.Module):
"""
Grouped GEMM (General Matrix Multiplication) module for efficient expert computation.
This module utilizes the grouped_gemm library (https://github.com/fanshiqing/grouped_gemm)
for optimized performance. If the grouped_gemm library is not installed, it gracefully
falls back to a sequential GEMM implementation, which may be slower but ensures
functionality.
Args:
in_features (`int`):
Number of input features.
out_features (`int`):
Number of output features.
groups (`int`):
Number of expert groups.
"""
def __init__(self, in_features, out_features, groups):
super().__init__()
self.in_features = in_features
self.out_features = out_features
self.groups = groups
self.weight = nn.Parameter(torch.empty(groups, in_features, out_features))
def forward(self, input, tokens_per_expert):
"""
Perform grouped matrix multiplication.
Args:
input (`torch.Tensor`):
Input tensor of shape (num_tokens, in_features).
tokens_per_expert (`torch.Tensor`):
Number of tokens assigned to each expert.
Returns:
torch.Tensor: Output tensor of shape (num_tokens, out_features).
"""
return sequential_experts_gemm(
input,
self.weight,
tokens_per_expert.cpu(),
)
class AriaGroupedExpertsMLP(nn.Module):
"""
Grouped MLP module for Mixture of Experts.
Args:
config (`AriaTextConfig`):
Configuration object for the model.
"""
def __init__(self, config: AriaTextConfig) -> None:
super().__init__()
self.config = config
self.fc1 = AriaGroupedExpertsGemm(config.hidden_size, config.intermediate_size * 2, config.moe_num_experts)
self.fc2 = AriaGroupedExpertsGemm(config.intermediate_size, config.hidden_size, config.moe_num_experts)
def forward(self, permuted_tokens, tokens_per_expert):
"""
Forward pass of the Grouped MLP.
Args:
permuted_tokens (torch.Tensor): Permuted input tokens.
tokens_per_expert (torch.Tensor): Number of tokens assigned to each expert.
Returns:
torch.Tensor: Output tensor after passing through the MLP.
"""
fc1_output = self.fc1(permuted_tokens, tokens_per_expert)
projection, gate = torch.chunk(fc1_output, 2, dim=-1)
fc1_output = nn.functional.silu(projection) * gate
fc2_output = self.fc2(fc1_output, tokens_per_expert)
return fc2_output
# Token permutation adapted from https://github.com/NVIDIA/Megatron-LM/blob/54f1f78529cbc2b9cddad313e7f9d96ac0420a27/megatron/core/transformer/moe/token_dispatcher.py#L291-L587
class AriaTextMoELayer(nn.Module):
"""
Aria Text Mixture of Experts (MoE) Layer.
This layer applies a gating mechanism to route input tokens to different experts.
Args:
config (`AriaTextConfig`):
Configuration object for the text component of the model.
"""
def __init__(self, config: AriaTextConfig):
super().__init__()
self.router = nn.Linear(config.hidden_size, config.moe_num_experts, bias=False)
self.experts = AriaGroupedExpertsMLP(config)
self.shared_experts = AriaSharedExpertsMLP(config)
self.config = config
def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
"""
Forward pass of the MoE Layer.
Args:
hidden_states (`torch.Tensor`):
Input tensor of shape (batch_size, sequence_length, hidden_size).
Returns:
torch.Tensor: Output tensor after passing through the MoE layer.
Process:
1. Route tokens to experts using the router.
2. Permute tokens based on routing decisions.
3. Process tokens through experts.
4. Unpermute and combine expert outputs.
5. Add shared expert output to the final result.
"""
original_shape = hidden_states.shape
hidden_states = hidden_states.view(-1, hidden_states.size(-1))
# Top K Routing
logits = self.router(hidden_states)
top_logits, top_indices = torch.topk(logits, k=self.config.moe_topk, dim=1)
scores = nn.functional.softmax(top_logits, dim=-1)
original_dtype = top_indices.dtype
tokens_per_expert = torch.histc(
top_indices.flatten().to(torch.float32),
bins=self.config.moe_num_experts,
min=0,
max=self.config.moe_num_experts - 1,
).to(original_dtype)
indices = top_indices
# Token permutation
flatten_indices = indices.view(-1)
sorted_indices = torch.argsort(flatten_indices)
permuted_tokens = hidden_states.index_select(0, sorted_indices // self.config.moe_topk)
# Process through experts
expert_output = self.experts(permuted_tokens, tokens_per_expert)
# Token unpermutation
unpermuted_tokens = torch.zeros(
(scores.shape[0] * self.config.moe_topk, expert_output.size(1)),
dtype=expert_output.dtype,
device=expert_output.device,
)
unpermuted_tokens.index_copy_(0, sorted_indices, expert_output)
unpermuted_tokens = unpermuted_tokens.view(-1, self.config.moe_topk, expert_output.size(1))
output = (unpermuted_tokens * scores.unsqueeze(-1)).sum(dim=1).view(original_shape)
# Add shared expert output
shared_expert_output = self.shared_experts(hidden_states.view(original_shape))
return output + shared_expert_output
class AriaTextAttention(LlamaAttention):
"""Multi-headed attention from 'Attention Is All You Need' paper"""
def __init__(self, config: AriaTextConfig, layer_idx: int):
super().__init__()
class AriaTextDecoderLayer(LlamaDecoderLayer):
"""
Aria Text Decoder Layer.
This class defines a single decoder layer in the language model, incorporating self-attention and Mixture of Experts (MoE) feed-forward network.
Args:
config (`AriaTextConfig`):
Configuration object for the text component of the model.
layer_idx (`int`):
Index of the layer.
"""
def __init__(self, config: AriaTextConfig, layer_idx: int):
super().__init__(self)
self.mlp = AriaTextMoELayer(config)
@auto_docstring
class AriaTextPreTrainedModel(PreTrainedModel):
config: AriaTextConfig
base_model_prefix = "model"
_no_split_modules = ["AriaTextDecoderLayer", "AriaGroupedExpertsGemm"]
supports_gradient_checkpointing = True
_skip_keys_device_placement = "past_key_values"
_supports_flash_attn = False
_supports_sdpa = True
_supports_attention_backend = True
_can_record_outputs = {
"hidden_states": AriaTextDecoderLayer,
"attentions": AriaTextAttention,
}
def _init_weights(self, module):
super()._init_weights(module)
if isinstance(module, AriaGroupedExpertsGemm):
module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
class AriaPreTrainedModel(LlamaPreTrainedModel):
config: AriaConfig
base_model_prefix = ""
_can_compile_fullgraph = False # MoE models don't work with torch.compile (dynamic slicing)
_supports_attention_backend = True
def _init_weights(self, module):
LlamaPreTrainedModel._init_weights(module)
if isinstance(module, AriaProjector):
nn.init.trunc_normal_(module.query, std=self.config.initializer_range)
class AriaTextModel(LlamaModel):
def __init__(self, config: AriaTextConfig):
super().__init__(config)
self.layers = nn.ModuleList(
[AriaTextDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
)
self.gradient_checkpointing = False
self.post_init()
class AriaTextForCausalLM(AriaTextPreTrainedModel, LlamaForCausalLM):
_tied_weights_keys = ["lm_head.weight"]
def __init__(self, config: AriaTextConfig):
super().__init__(config)
self.model = AriaTextModel(config)
self.vocab_size = config.vocab_size
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
# Initialize weights and apply final processing
self.post_init()
@auto_docstring
def forward(self, **super_kwargs):
super().forward(self, **super_kwargs)
class AriaCausalLMOutputWithPast(LlavaCausalLMOutputWithPast):
pass
class AriaModelOutputWithPast(LlavaModelOutputWithPast):
pass
class AriaModel(LlavaModel):
def __init__(self, config: AriaConfig):
super().__init__(config)
self.multi_modal_projector = AriaProjector(config)
def _create_patch_attention_mask(self, pixel_mask):
if pixel_mask is None:
return None
patches_subgrid = pixel_mask.unfold(
dimension=1,
size=self.vision_tower.config.patch_size,
step=self.vision_tower.config.patch_size,
)
patches_subgrid = patches_subgrid.unfold(
dimension=2,
size=self.vision_tower.config.patch_size,
step=self.vision_tower.config.patch_size,
)
return (patches_subgrid.sum(dim=(-1, -2)) > 0).bool()
def get_image_features(
self,
pixel_values: torch.FloatTensor,
pixel_mask: Optional[torch.FloatTensor] = None,
vision_feature_layer: int = -1,
):
"""
Obtains image last hidden states from the vision tower and apply multimodal projection.
Args:
pixel_values (`torch.FloatTensor]` of shape `(batch_size, channels, height, width)`):
The tensors corresponding to the input images.
pixel_mask (`torch.FloatTensor]`, *optional*):
The tensors corresponding to the input image mask.
vision_feature_layer (`Union[int, list[int]]`, *optional*):
The index of the layer to select the vision feature. If multiple indices are provided,
the vision feature of the corresponding indices will be concatenated to form the
vision features.
Returns:
image_features (`torch.Tensor`): Image feature tensor of shape `(num_images, image_length, embed_dim)`).
"""
vision_feature_layer = (
vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer
)
patch_attention_mask = self._create_patch_attention_mask(pixel_mask)
image_outputs = self.vision_tower(
pixel_values, patch_attention_mask=patch_attention_mask, output_hidden_states=True
)
image_attn_mask = None
if patch_attention_mask is not None:
flattened_mask = patch_attention_mask.flatten(1)
image_attn_mask = torch.logical_not(flattened_mask)
selected_image_feature = image_outputs.hidden_states[vision_feature_layer]
image_features = self.multi_modal_projector(selected_image_feature, attn_mask=image_attn_mask)
return image_features
def forward(
self,
input_ids: torch.LongTensor = None,
pixel_values: torch.FloatTensor = None,
pixel_mask: torch.LongTensor = None,
attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None,
past_key_values: Optional[Cache] = 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,
cache_position: Optional[torch.LongTensor] = None,
**kwargs: Unpack[FlashAttentionKwargs],
) -> Union[tuple, AriaModelOutputWithPast]:
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 inputs_embeds is None:
inputs_embeds = self.get_input_embeddings()(input_ids)
# 2. Merge text and images
if pixel_values is not None and inputs_embeds.shape[1] != 1:
if input_ids is None:
special_image_mask = inputs_embeds == self.get_input_embeddings()(
torch.tensor(self.config.image_token_id, dtype=torch.long, device=inputs_embeds.device)
)
special_image_mask = special_image_mask.all(-1)
else:
special_image_mask = input_ids == self.config.image_token_id
n_image_tokens = (special_image_mask).sum(dim=1).sum(dim=0)
special_image_mask = special_image_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
image_features = self.get_image_features(
pixel_values=pixel_values,
pixel_mask=pixel_mask,
vision_feature_layer=self.config.vision_feature_layer,
)
n_images, n_features_per_image = image_features.shape[0], image_features.shape[1]
n_image_features = n_images * n_features_per_image
if n_image_tokens != n_image_features:
raise ValueError(
f"Image features and image tokens do not match: tokens: {n_image_tokens}, features {n_image_features}"
)
image_features = image_features.to(inputs_embeds.device, inputs_embeds.dtype)
inputs_embeds = inputs_embeds.masked_scatter(special_image_mask, image_features)
outputs = self.language_model(
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=True,
cache_position=cache_position,
**kwargs,
)
return AriaModelOutputWithPast(
last_hidden_state=outputs.last_hidden_state,
past_key_values=outputs.past_key_values if use_cache else None,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
image_hidden_states=image_features if pixel_values is not None else None,
)
@auto_docstring(
custom_intro="""
Aria model for conditional generation tasks.
This model combines a vision tower, a multi-modal projector, and a language model
to perform tasks that involve both image and text inputs.
"""
)
class AriaForConditionalGeneration(LlavaForConditionalGeneration):
def get_image_features(
self,
pixel_values: torch.FloatTensor,
pixel_mask: Optional[torch.FloatTensor] = None,
vision_feature_layer: int = -1,
):
return self.model.get_image_features(
pixel_values=pixel_values,
pixel_mask=pixel_mask,
vision_feature_layer=vision_feature_layer,
)
@can_return_tuple
@auto_docstring
def forward(
self,
input_ids: torch.LongTensor = None,
pixel_values: torch.FloatTensor = None,
pixel_mask: torch.LongTensor = None,
attention_mask: Optional[torch.Tensor] = None,
position_ids: Optional[torch.LongTensor] = None,
past_key_values: Optional[Cache] = 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,
logits_to_keep: Union[int, torch.Tensor] = 0,
cache_position: Optional[torch.LongTensor] = None,
**kwargs: Unpack[TransformersKwargs],
) -> Union[tuple, AriaCausalLMOutputWithPast]:
r"""
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 `model.image_token_id` (where `model` is your instance of `AriaForConditionalGeneration`).
Tokens with indices set to `model.image_token_id` are ignored (masked), the loss is only
computed for the tokens with labels in `[0, ..., config.vocab_size]`.
Example:
```python
>>> import requests
>>> import torch
>>> from PIL import Image
>>> from io import BytesIO
>>> from transformers import AutoProcessor, AutoModel
>>> from transformers.image_utils import load_image
>>> # Note that passing the image urls (instead of the actual pil images) to the processor is also possible
>>> image1 = load_image("https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg")
>>> image2 = load_image("https://cdn.britannica.com/59/94459-050-DBA42467/Skyline-Chicago.jpg")
>>> image3 = load_image("https://cdn.britannica.com/68/170868-050-8DDE8263/Golden-Gate-Bridge-San-Francisco.jpg")
>>> processor = AutoProcessor.from_pretrained("Rhymes-AI/Aria")
>>> model = AutoModel.from_pretrained("Rhymes-AI/Aria", torch_dtype=torch.bfloat16, device_map="auto")
>>> # Create inputs
>>> messages = [
... {
... "role": "user",
... "content": [
... {"type": "image"},
... {"type": "text", "text": "In this image, we can see the city of New York, and more specifically the Statue of Liberty."},
... {"type": "image"},
... {"type": "text", "text": "What can we see in this image?"},
... ]
... },
... {
... "role": "user",
... "content": [
... {"type": "image"},
... {"type": "text", "text": "In which city is that bridge located?"},
... ]
... }
... ]
>>> prompts = [processor.apply_chat_template([message], add_generation_prompt=True) for message in messages]
>>> images = [[image1, image2], [image3]]
>>> inputs = processor(text=prompts, images=images, padding=True, return_tensors="pt").to(model.device)
>>> # Generate
>>> generated_ids = model.generate(**inputs, max_new_tokens=256)
>>> generated_texts = processor.batch_decode(generated_ids, skip_special_tokens=True)
>>> print(generated_texts[0])
Assistant: There are buildings, trees, lights, and water visible in this image.
>>> print(generated_texts[1])
Assistant: The bridge is in San Francisco.
```"""
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
outputs = self.model(
input_ids=input_ids,
pixel_values=pixel_values,
pixel_mask=pixel_mask,
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,
cache_position=cache_position,
**kwargs,
)
hidden_states = outputs[0]
# Only compute necessary logits, and do not upcast them to float if we are not computing the loss
slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
logits = self.lm_head(hidden_states[:, slice_indices, :])
loss = None
if labels is not None:
loss = self.loss_function(
logits=logits, labels=labels, vocab_size=self.config.text_config.vocab_size, **kwargs
)
return AriaCausalLMOutputWithPast(
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,
inputs_embeds=None,
pixel_values=None,
pixel_mask=None,
attention_mask=None,
cache_position=None,
logits_to_keep=None,
**kwargs,
):
model_inputs = super().prepare_inputs_for_generation(
input_ids,
past_key_values=past_key_values,
inputs_embeds=inputs_embeds,
attention_mask=attention_mask,
cache_position=cache_position,
logits_to_keep=logits_to_keep,
**kwargs,
)
if cache_position[0] == 0:
# If we're in cached decoding stage, pixel values should be None because input ids do not contain special image token anymore
# Otherwise we need pixel values to be passed to model
model_inputs["pixel_values"] = pixel_values
model_inputs["pixel_mask"] = pixel_mask
return model_inputs
__all__ = [
"AriaConfig",
"AriaTextConfig",
"AriaImageProcessor",
"AriaProcessor",
"AriaForConditionalGeneration",
"AriaPreTrainedModel",
"AriaTextPreTrainedModel",
"AriaTextModel",
"AriaModel",
"AriaTextForCausalLM",
]