| ---
|
| license: apache-2.0
|
| tags:
|
| - pytorch
|
| - deit
|
| - generation
|
| - wip
|
| - prototype
|
| ---
|
|
|
| # DeiT experiments: Generation
|
|
|
| ## Overview
|
|
|
| A research-oriented **DeiT** prototype targeting **Generation**. The included **nano** setup documents defaults and file formats without presenting unverified performance numbers.
|
|
|
| ## Quick Check
|
|
|
| ```bash
|
| python predict.py --help
|
| ```
|
|
|
| Inspect the script's `__main__` block for its generated smoke-test example. Because this is a custom implementation, generic automatic loading APIs require an explicit adapter before use.
|
|
|
| ## Contact
|
|
|
| Questions or corrections: open a discussion in this repository.
|
|
|
| ## Training Details
|
|
|
| | Item | Value |
|
| |---|---|
|
| | Framework | PyTorch |
|
| | Precision | fp32 |
|
| | Init scheme | xavier uniform |
|
| | Hardware | a single consumer GPU is enough |
|
| | Typical run | minutes, not hours |
|
|
|
| ## Limitations
|
|
|
| The initialization checkpoint has not been trained or audited for robustness, fairness, or domain transfer. The implementation should be treated as an experimental starting point. Results from a future trained checkpoint must be documented separately from the defaults shipped here.
|
|
|
| ## Default Experiment Recipe
|
|
|
| The included configuration uses **sgd** with a **constant warmup** schedule. These are starting values in the script, not evidence of a completed run. For a meaningful evaluation, train all baselines with the same data exposure, tuning budget, and random seeds.
|
|
|
| ## Architecture
|
|
|
| | Item | Value |
|
| |---|---|
|
| | Architecture | DeiT |
|
| | Scale | nano |
|
| | Attention | multi query |
|
| | Fusion | tucker |
|
| | Activation | swish |
|
| | Normalization | scalenorm |
|
|
|
| ## Repository Status
|
|
|
| 1. The Python file contains the model and runnable example or training entry point.
|
| 2. `config.json` records the generated architecture settings.
|
| 3. `training_args.json` records the default experiment recipe.
|
| 4. `model.safetensors` is a valid initialization checkpoint for smoke tests; it is **not** presented as a trained benchmark checkpoint.
|
| 5. No benchmark score is claimed in this repository.
|
|
|
| ## Files
|
|
|
| | File | Purpose |
|
| |---|---|
|
| | `predict.py` | primary artifact |
|
| | `README.md` | this documentation |
|
| | `config.json` | architecture configuration |
|
| | `training_args.json` | default experiment settings |
|
| | `model.safetensors` | initialization checkpoint |
|
|
|
| ## License
|
|
|
| Released under **apache-2.0**. Review the source-data terms separately when this repository is used with external datasets.
|
|
|
