Testing-model / README.md
KhangTruong's picture
feat(release): final training report and checkpoints for KhangTruong/Testing-model (v1)
1d7c979 verified
|
Raw History Blame Contribute Delete
2.02 kB
---
license: apache-2.0
tags:
- pytorch
- sid-unet
- image-segmentation
- synthetic-image-detection
- diffusion-detection
pipeline_tag: image-segmentation
library_name: sid-unet
---
# Testing-model
Official repository for **Testing-model** checkpoints, evaluated on AI-generated image detection and segmentation tasks.
## πŸ“Œ Available Versions
| Version | Best Metric Score | Best Epoch | Updated |
| :--- | :--- | :--- | :--- |
| `v1` | 0.6 | 1 | 2026-10-28 |
- **Active Version:** `v1`
- **Best Validation Score:** `0.6` (Epoch `1`)
- **Total Training Epochs:** `22`
---
## πŸš€ Quickstart & Checkpoint Usage
### 1. Pull checkpoints using `sid-pull`
To download the latest checkpoint directly to your local workspace:
```bash
sid-pull --name KhangTruong/Testing-model
```
To pull a specific version:
```bash
sid-pull --name KhangTruong/Testing-model --version v1
```
### 2. Resume or Fine-tune with `sid-train`
Once downloaded, resume training seamlessly:
```bash
sid-train --config configs/experiments/diffusion_diff_minimized/default.yaml --override training.epochs=40
```
Or specify explicit checkpoint path:
```bash
sid-train --resume checkpoints/checkpoint_latest.pt
```
### 3. Push new checkpoints using `sid-push`
```bash
sid-push --name KhangTruong/Testing-model --version v2
```
---
## πŸ“‚ Repository Layout
```
KhangTruong/Testing-model/
β”œβ”€β”€ README.md # Model Card & Documentation
β”œβ”€β”€ config.yaml # Model & Training Configuration
β”œβ”€β”€ manifest.json # Catalog of all versions & metrics
β”œβ”€β”€ checkpoints/ # Active / Latest Checkpoints
β”‚ β”œβ”€β”€ checkpoint_best.pt
β”‚ β”œβ”€β”€ checkpoint_latest.pt
β”‚ └── checkpoint_periodic.pt
β”œβ”€β”€ reports/ # Evaluation reports & plots
β”‚ β”œβ”€β”€ training_curves.png
β”‚ └── test_evaluation_report.json
└── versions/ # Version Archive
└── v1/
β”œβ”€β”€ checkpoints/
└── reports/
```