# deeptb-helper

> Use this skill to set up and run DeePTB (both SKTB and E3TB workflows).

- Author: Siyu Liu
- Repository: IntelligentMat/Materials-Science-Skills-For-LLM
- Version: 20251221161123
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-07
- Source: https://github.com/IntelligentMat/Materials-Science-Skills-For-LLM
- Web: https://mule.run/skillshub/@@IntelligentMat/Materials-Science-Skills-For-LLM~deeptb-helper:20251221161123

---

---
name: deeptb-helper
description: Operate the DeePTB Python package (Slater-Koster and E3 TB models): install via uv/pip, prepare datasets (info.json, eigenvalues/hamiltonian labels), generate config templates, train with dptb train, run inference/plots with dptb run, and use bundled examples/troubleshooting tips. Trigger when asked to install, configure, train, run, or debug DeePTB.
---

# DeePTB Helper

Use this skill to set up and run DeePTB (both SKTB and E3TB workflows).

## Quick Start Flow
1) **Install** (CPU/GPU) using `references/install.md`.
2) **Inspect examples** in `examples/hBN` (SKTB) or `examples/e3` (E3) for ready-to-run configs.
3) **Prepare data** and `info.json` per `references/datasets.md`.
4) **Generate a config template** with `dptb config -tr [-sk|-skenv|-e3] <path/input.json>`.
5) **Edit key sections**: `common_options`, `model_options` (match SK/E3), `train_options`, and `data_options`.
6) **Train**: `dptb train <input.json> [-o out_dir] [-i|-r ckpt]`.
7) **Evaluate/plot**: `dptb run <run.json> -i <ckpt> -o <out_dir>`; use `dptb bond <structure>` to pick cutoffs.

## Essential Commands
- `dptb --help` for CLI overview.
- `dptb config -h` to list template modes.
- `dptb bond <structure> [-c cutoff]` to inspect neighbor distances (helps choose `r_max`/`rs`).
- `dptb train <input> -o <out>` to fit; `-i`/`-r` continues from checkpoints.
- `dptb run <run.json> -i <ckpt> -o <out>` to predict/plot bands or generate TBtrans inputs.
- `uv run dptb ...` if using the uv-managed venv.

## When to Load References
- Need install variants or env activation → read `references/install.md`.
- Need dataset layout, `info.json`, or label formats → read `references/datasets.md`.
- Need CLI examples/templates for SKTB vs E3TB → read `references/cli-cheatsheet.md`.

## Tips and Checks
- Match `basis` in `run` scripts to training `basis`.
- For GPU installs, ensure CUDA version matches torch_scatter wheel.
- Keep `r_max` / `rs` consistent between data prep and model options.
- If bands look misaligned, verify `bandinfo` window and that core bands are excluded.
- Unit tests: `uv run pytest ./dptb/tests` (optional sanity check after install).