Improve model card: Add pipeline tag, library name, and relevant tags

This PR enhances the model card by adding crucial metadata to improve its discoverability and utility on the Hugging Face Hub:

* **`pipeline_tag: image-to-image`**: The model performs various image restoration tasks (deraining, dehazing, denoising), which fall under the `image-to-image` pipeline category. This will ensure the model appears in relevant task filters.
* **`library_name: pytorch`**: The GitHub repository explicitly states it's a "PyTorch implementation" and provides PyTorch-based training and testing scripts. Adding this tag enables an automated PyTorch usage snippet on the Hub, making it easier for users to get started.
* **`tags`**: Specific tags such as `image-restoration`, `deraining`, `dehazing`, and `denoising` have been added. These reflect the core functionalities of the MIR-L model as described in the paper abstract, further improving searchability for specific image degradation tasks.

The existing arXiv paper link has been retained as per instructions, and no sample usage code has been added since it wasn't present in the original GitHub README.

README.md

@@ -1,19 +1,26 @@
----
-license: mit
----
-
-# MIR-L
-
-[![paper](https://img.shields.io/badge/arXiv-Paper-<COLOR>.svg)](https://arxiv.org/abs/2510.14463)
-
-This repository provides the pre-trained checkpoints of the MIR-L model proposed in the paper titled **Pruning Overparameterized Multi-Task Networks for Degraded Web Image Restoration**.
-
-> **Abstract:** Image quality is a critical factor in delivering visually appealing content on web platforms. However, images often suffer from degradation due to lossy operations applied by online social networks (OSNs), negatively affecting user experience. Image restoration is the process of recovering a clean high-quality image from a given degraded input. Recently, multi-task (all-in-one) image restoration models have gained significant attention, due to their ability to simultaneously handle different types of image degradations. However, these models often come with an excessively high number of trainable parameters, making them computationally inefficient. In this paper, we propose a strategy for compressing multi-task image restoration models. We aim to discover highly sparse subnetworks within overparameterized deep models that can match or even surpass the performance of their dense counterparts. The proposed model, namely MIR-L, utilizes an iterative pruning strategy that removes low-magnitude weights across multiple rounds, while resetting the remaining weights to their original initialization. This iterative process is important for the multi-task image restoration model’s optimization, effectively uncovering “winning tickets” that maintain or exceed state-of-the-art performance at high sparsity levels. Experimental evaluation on benchmark datasets for the deraining, dehazing, and denoising tasks shows that MIR-L retains only 10% of the trainable parameters while maintaining high image restoration performance.
-
----
-
-## Installation, Datasets, and Usage
-
-Check out the [official repository](https://github.com/Thomkat/MIR-L) for details.
-
+---
+license: mit
+pipeline_tag: image-to-image
+library_name: pytorch
+tags:
+  - image-restoration
+  - deraining
+  - dehazing
+  - denoising
+---
+
+# MIR-L
+
+[![paper](https://img.shields.io/badge/arXiv-Paper-<COLOR>.svg)](https://arxiv.org/abs/2510.14463)
+
+This repository provides the pre-trained checkpoints of the MIR-L model proposed in the paper titled **Pruning Overparameterized Multi-Task Networks for Degraded Web Image Restoration**.
+
+> **Abstract:** Image quality is a critical factor in delivering visually appealing content on web platforms. However, images often suffer from degradation due to lossy operations applied by online social networks (OSNs), negatively affecting user experience. Image restoration is the process of recovering a clean high-quality image from a given degraded input. Recently, multi-task (all-in-one) image restoration models have gained significant attention, due to their ability to simultaneously handle different types of image degradations. However, these models often come with an excessively high number of trainable parameters, making them computationally inefficient. In this paper, we propose a strategy for compressing multi-task image restoration models. We aim to discover highly sparse subnetworks within overparameterized deep models that can match or even surpass the performance of their dense counterparts. The proposed model, namely MIR-L, utilizes an iterative pruning strategy that removes low-magnitude weights across multiple rounds, while resetting the remaining weights to their original initialization. This iterative process is important for the multi-task image restoration model’s optimization, effectively uncovering “winning tickets” that maintain or exceed state-of-the-art performance at high sparsity levels. Experimental evaluation on benchmark datasets for the deraining, dehazing, and denoising tasks shows that MIR-L retains only 10% of the trainable parameters while maintaining high image restoration performance.
+
+---
+
+## Installation, Datasets, and Usage
+
+Check out the [official repository](https://github.com/Thomkat/MIR-L) for details.
+
 ---