configuration_internvl_chat.py

# --------------------------------------------------------
# InternVL
# Copyright (c) 2024 OpenGVLab
# Licensed under The MIT License [see LICENSE for details]
# --------------------------------------------------------

import copy

import copy
import os
import sys

from transformers import AutoConfig, LlamaConfig, Qwen2Config
from transformers.configuration_utils import PretrainedConfig
from transformers.utils import logging

# Handle both relative and absolute imports for dynamic module loading
import importlib
import importlib.util

try:
    from .configuration_intern_vit import InternVisionConfig
except (ImportError, ValueError) as e:
    # When loaded as a dynamic module by transformers, try to find already-loaded config first
    InternVisionConfig = None
    
    # Check if config module is already loaded in sys.modules
    for module_name in list(sys.modules.keys()):
        if 'configuration_intern_vit' in module_name:
            try:
                InternVisionConfig = sys.modules[module_name].InternVisionConfig
                break
            except AttributeError:
                pass
    
    # If not found, load from file
    if InternVisionConfig is None:
        current_dir = os.path.dirname(os.path.abspath(__file__))
        config_file = os.path.join(current_dir, "configuration_intern_vit.py")
        spec = importlib.util.spec_from_file_location("configuration_intern_vit", config_file)
        config_module = importlib.util.module_from_spec(spec)
        spec.loader.exec_module(config_module)
        InternVisionConfig = config_module.InternVisionConfig

logger = logging.get_logger(__name__)


class InternVLChatConfig(PretrainedConfig):
    model_type = 'internvl_chat'
    is_composition = True

    def __init__(
            self,
            vision_config=None,
            llm_config=None,
            use_backbone_lora=0,
            use_llm_lora=0,
            select_layer=-1,
            force_image_size=None,
            downsample_ratio=0.5,
            template=None,
            dynamic_image_size=False,
            use_thumbnail=False,
            ps_version='v1',
            min_dynamic_patch=1,
            max_dynamic_patch=6,
            onnx_path=None,
            vision_onnx_file=None,
            **kwargs):
        super().__init__(**kwargs)

        if vision_config is None:
            vision_config = {'architectures': ['InternVisionModel']}
            logger.info('vision_config is None. Initializing the InternVisionConfig with default values.')

        if llm_config is None:
            llm_config = {'architectures': ['Qwen2ForCausalLM']}
            logger.info('llm_config is None. Initializing the LlamaConfig config with default values (`LlamaConfig`).')

        self.vision_config = InternVisionConfig(**vision_config)
        if llm_config.get('architectures')[0] == 'LlamaForCausalLM':
            self.llm_config = LlamaConfig(**llm_config)
        elif llm_config.get('architectures')[0] == 'Qwen2ForCausalLM':
            self.llm_config = Qwen2Config(**llm_config)
        else:
            raise ValueError('Unsupported architecture: {}'.format(llm_config.get('architectures')[0]))
        self.use_backbone_lora = use_backbone_lora
        self.use_llm_lora = use_llm_lora
        self.select_layer = select_layer
        self.force_image_size = force_image_size
        self.downsample_ratio = downsample_ratio
        self.template = template
        self.dynamic_image_size = dynamic_image_size
        self.use_thumbnail = use_thumbnail
        self.ps_version = ps_version  # pixel shuffle version
        self.min_dynamic_patch = min_dynamic_patch
        self.max_dynamic_patch = max_dynamic_patch
        self.onnx_path = onnx_path
        self.vision_onnx_file = vision_onnx_file
        # By default, we use tie_word_embeddings=False for models of all sizes.
        self.tie_word_embeddings = self.llm_config.tie_word_embeddings

        logger.info(f'vision_select_layer: {self.select_layer}')
        logger.info(f'ps_version: {self.ps_version}')
        logger.info(f'min_dynamic_patch: {self.min_dynamic_patch}')
        logger.info(f'max_dynamic_patch: {self.max_dynamic_patch}')

    def to_dict(self):
        """
        Serializes this instance to a Python dictionary. Override the default [`~PretrainedConfig.to_dict`].

        Returns:
            `Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance,
        """
        output = copy.deepcopy(self.__dict__)
        output['vision_config'] = self.vision_config.to_dict()
        output['llm_config'] = self.llm_config.to_dict()
        output['model_type'] = self.__class__.model_type
        output['use_backbone_lora'] = self.use_backbone_lora
        output['use_llm_lora'] = self.use_llm_lora
        output['select_layer'] = self.select_layer
        output['force_image_size'] = self.force_image_size
        output['downsample_ratio'] = self.downsample_ratio
        output['template'] = self.template
        output['dynamic_image_size'] = self.dynamic_image_size
        output['use_thumbnail'] = self.use_thumbnail
        output['ps_version'] = self.ps_version
        output['min_dynamic_patch'] = self.min_dynamic_patch
        output['max_dynamic_patch'] = self.max_dynamic_patch
        output['onnx_path'] = self.onnx_path
        output['vision_onnx_file'] = self.vision_onnx_file

        return output