README.md

metadata

title: Transfer learning results (%)
license: mit
task_categories:
  - image-classification
tags:
  - transfer-learning
  - resnet
  - vit
  - vit-hybrid
  - swin-transformer

Transfer Learning Results

This repository contains the weights and experimental results of transfer learning fine-tuning using several mainstream vision backbones. The method was evaluated based on Cross Entropy.

📂 Directory Structure

.
├── ResNet-34/
│   ├── config.json              # Model configuration parameters
│   ├── model.safetensors        # Model weights file
│   └── training_args.bin        # Training hyperparameters and arguments
├── Swin Transformer/
│   ├── config.json
│   ├── model.safetensors
│   └── training_args.bin
├── ViT/
│   ├── config.json
│   ├── model.safetensors
│   └── training_args.bin
└── ViT-H/
    ├── config.json
    ├── model.safetensors
    └── training_args.bin

🚀 Usage

You can download the model weights for a specific continual learning method using the huggingface_hub SDK:

from huggingface_hub import hf_hub_download

# Example: Download Swin Transformer model weights
REPO_ID = "Dataset-and-Model/transfer-learning"
config_path = hf_hub_download(
    repo_id=REPO_ID,
    filename="Swin Transformer/config.json",
    repo_type="model"
)
model_weight_path = hf_hub_download(
    repo_id=REPO_ID,
    filename="Swin Transformer/model.safetensors",
    repo_type="model"
)
training_args_path = hf_hub_download(
    repo_id=REPO_ID,
    filename="Swin Transformer/training_args.bin",
    repo_type="model"
)

print(f"Weights successfully downloaded to: {model_weight_path}")

🏗️ Base Models

The fine-tuning and continual learning experiments in this study are built upon the following pre-trained models provided by Hugging Face:

• ResNet-34 (e.g., microsoft/resnet-34)
• Vision Transformer (ViT) (e.g., google/vit-large-patch32-384)
• ViT Hybrid (e.g., google/vit-hybrid-base-bit-384)
• SWIN Transformer (e.g., microsoft/swin-large-patch4-window12-384)

📜 License

These fine-tuned model weights are derived from the open-source architectures mentioned above and are released under the MIT License. You are free to use, modify, and distribute these weights in both academic research and commercial applications, provided that proper attribution to the original authors is maintained.