Delete modeling.py

modeling.py

@@ -1,129 +0,0 @@
-#
-#  Copyright (c) 2025
-#  Minh NGUYEN <vnguyen9@lakeheadu.ca>
-#
-import logging
-import torch
-import torch.nn as nn
-import torch.nn.init as init
-
-from transformers import PreTrainedModel
-from typing import Callable, Any, Optional
-from typing_extensions import Self
-
-from collections import OrderedDict
-
-from .configuration import AlexNetConfig
-
-logger = logging.getLogger(__name__)
-
-
-class AlexNet(PreTrainedModel):
-    """The Alex Network class."""
-
-    config_class = AlexNetConfig
-
-    def __init__(self, config: Optional[AlexNetConfig] = None):
-        config = config or AlexNetConfig()
-
-        super().__init__(config)
-
-        self.config = config
-
-        self.feature_extractor = nn.Sequential(
-            OrderedDict(
-                [
-                    ("conv1", nn.Conv2d(in_channels=3, out_channels=64, kernel_size=11, stride=4, padding=2)),
-                    ("relu1", nn.ReLU(inplace=True)),
-                    ("maxpool1", nn.MaxPool2d(kernel_size=3, stride=2)),
-
-                    ("conv2", nn.Conv2d(in_channels=64, out_channels=192, kernel_size=5, padding=2, bias=True)),
-                    ("relu2", nn.ReLU(inplace=True)),
-                    ("maxpool2", nn.MaxPool2d(kernel_size=3, stride=2)),
-
-                    ("conv3", nn.Conv2d(in_channels=192, out_channels=384, kernel_size=3, padding=1)),
-                    ("reul3", nn.ReLU(inplace=True)),
-
-                    ("conv4", nn.Conv2d(in_channels=384, out_channels=256, kernel_size=3, padding=1)),
-                    ("relu4", nn.ReLU(inplace=True)),
-
-                    ("conv5", nn.Conv2d(in_channels=256, out_channels=256, kernel_size=3, padding=1)),
-                    ("relu5", nn.ReLU(inplace=True)),
-                    ("maxpool3", nn.MaxPool2d(kernel_size=3, stride=2)),
-                ]
-            )
-        )
-
-        self.avgpool = nn.AdaptiveAvgPool2d(output_size=(6, 6))
-
-        self.head = nn.Sequential(
-            OrderedDict(
-                [
-                    ("dropout1", nn.Dropout(p=0.5)),
-                    ("linear1", nn.Linear(in_features=256 * 6 * 6, out_features=4096, bias=True)),
-                    ("relu1", nn.ReLU(inplace=True)),
-                    ("dropout2", nn.Dropout(p=0.5)),
-                    ("linear2", nn.Linear(in_features=4096, out_features=4096, bias=True)),
-                    ("relu2", nn.ReLU(inplace=True)),
-                    ("linear3", nn.Linear(in_features=4096, out_features=1000, bias=True)),
-                ]
-            )
-        )
-
-    def forward(self, x: torch.Tensor) -> torch.Tensor:
-        x = self.feature_extractor(x)
-        x = self.avgpool(x)
-        x = torch.flatten(x, start_dim=1, end_dim=-1)
-        x = self.head(x)
-
-        return x
-
-    def init_weights_(self, fn: Callable[[nn.Module], None]) -> Self:
-        self.apply(fn)
-
-        return self
-
-
-@torch.no_grad()
-def init_weights(m: nn.Module):
-    """Initialize weight for the module."""
-    try:
-        if isinstance(m, nn.Linear):
-            init.xavier_uniform_(m.weight)
-            if m.bias is not None:
-                init.zeros_(m.bias)
-        elif isinstance(m, nn.Conv2d):
-            init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
-            if m.bias is not None:
-                init.zeros_(m.bias)
-    except Exception as e:
-        logger.error(f"Error initializing weight module {m}.")
-
-
-def build_alexnet(*, weight_path: Optional[str] = None, **model_kwargs: Any) -> AlexNet:
-    """AlexNet model architecture from `One weird trick for parallelizing convolutional neural netwroks <https://arxiv.org/abs/1404.5997>`__.
-
-    .. note::
-        AlexNet was originally introduced in the `Image Classification` with Deep Convolutional Neural Network
-        <https://papers.nips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html>`__
-        paper. Our implementation is based instead on the `One weird trick` paper above.
-
-    Args:
-        weight_path (str): path to saved weights.
-        **model_kwargs: parameters passed to model.
-    """
-
-    model = AlexNet(**model_kwargs)
-
-    model.init_weights_(init_weights)
-    logger.info("Initialized random weights.")
-
-    if weight_path:
-        try:
-            state_dict = torch.load(weight_path)
-            model.load_state_dict(state_dict=state_dict, strict=False, assign=True)
-            logger.info(f"Loaded state dict from {weight_path!r}.")
-        except Exception as e:
-            logger.error(f"Error loading state dict: {e}")
-
-    return model