|
Download SOURCE_PACKAGE_NOTES.md from zerdovzad/WarpStateAI: direct link, hf CLI and curl.
- Browser
- Download file 1.52 kB
-
https://huggingface.co/zerdovzad/WarpStateAI/resolve/main/SOURCE_PACKAGE_NOTES.md
- Command line
-
hf download hf://zerdovzad/WarpStateAI/SOURCE_PACKAGE_NOTES.md
-
curl -L -o SOURCE_PACKAGE_NOTES.md https://huggingface.co/zerdovzad/WarpStateAI/resolve/main/SOURCE_PACKAGE_NOTES.md
1.52 kB
| # WarpState β assembled source bundle | |
| This bundle combines the **three source archives you provided** with the newly designed README and SVG assets. Existing source files were preserved byte-for-byte. No Python implementation was rewritten. Python bytecode caches (`__pycache__`, `.pyc`) were omitted. | |
| ## Included | |
| - `warpstate/model.py`, `warpstate/config.py`, `warpstate/__init__.py` β model implementation | |
| - `run_40b.py` β pretraining | |
| - `posttrain.py` β supervised fine-tuning | |
| - `dpo.py` β preference optimization | |
| - `chat.py` β chat inference/probe | |
| - `ops/*.py` and `ops/*.sh` β execution and training operations | |
| - `test_*.py` β tests | |
| - `tokenizer.model` β supplied tokenizer | |
| - `README.md` and `assets/` β updated GitHub presentation | |
| ## Not supplied in the uploaded archives | |
| - Trained model checkpoints / weights | |
| - Training datasets | |
| - `configs/` directory, including `configs/warpstate_2b.json` referenced by supplied `PACKAGE_MANIFEST.json` and training documents | |
| - Explicit license file | |
| **Note:** This is a *complete assembly of the available source*, not a turnkey pretrained model. Before publishing, review all source and documentation, choose a license, restore valid configs, and independently validate environment-specific commands. Scripts with old server-specific paths may require adaptation. Do not upload private secrets or checkpoints. | |
| Original archives: | |
| - `drive-download-20261026T200043Z-1-001.zip` | |
| - `warpstate-20261026T200055Z-1-001.zip` | |
| - `ops-20261026T200120Z-1-001.zip` | |