configuration_painter.py

"""
https://huggingface.co/docs/transformers/en/custom_models

Yongming
2025.10.23

Learn hugging face Transformers Custom Models - VLM version for vLLM compatibility.

"""

from transformers import PretrainedConfig

class PainterConfig(PretrainedConfig):
    model_type = "painter"

    def __init__(
        self,
        # Vision parameters
        input_channels: int = 3,
        hidden_size: int = 512,
        vision_hidden_size: int = 256,
        # Language model parameters
        vocab_size: int = 32000,
        hidden_act: str = "gelu",
        intermediate_size: int = 2048,
        max_position_embeddings: int = 2048,
        num_attention_heads: int = 8,
        num_hidden_layers: int = 6,
        # VLM specific parameters
        image_size: int = 224,
        patch_size: int = 16,
        num_patches: int = 196,  # (224/16)^2
        **kwargs,
    ):
        super().__init__(**kwargs)
        # Vision parameters
        self.input_channels = input_channels
        self.hidden_size = hidden_size
        self.vision_hidden_size = vision_hidden_size
        self.image_size = image_size
        self.patch_size = patch_size
        self.num_patches = num_patches
        
        # Language model parameters
        self.vocab_size = vocab_size
        self.hidden_act = hidden_act
        self.intermediate_size = intermediate_size
        self.max_position_embeddings = max_position_embeddings
        self.num_attention_heads = num_attention_heads
        self.num_hidden_layers = num_hidden_layers