# Nanogpt Training

Train GPT-2 scale models (~124M parameters) efficiently on a single GPU.

- Repository: https://github.com/benchflow-ai/skillsbench (1888 stars)
- Category: Data and ML
- License: Apache-2.0
- Last updated: 2026-10-24
- Runs scripts: no, instructions only

## Install

Claude Code: `npx skills add benchflow-ai/skillsbench --skill nanogpt-training -g -a claude-code -y`
Codex: `npx skills add benchflow-ai/skillsbench --skill nanogpt-training -g -a codex -y`
Cursor: `npx skills add benchflow-ai/skillsbench --skill nanogpt-training -g -a cursor -y`
Gemini CLI: `npx skills add benchflow-ai/skillsbench --skill nanogpt-training -g -a gemini-cli -y`

## Description

Train GPT-2 scale models (~124M parameters) efficiently on a single GPU. Covers GPT-124M architecture, tokenized dataset loading (e.g., HuggingFace Hub shards), modern optimizers (Muon, AdamW), mixed precision training, and training loop implementation.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# NanoGPT Training

## Overview

Training GPT-2 scale models (~124M parameters) efficiently on a single GPU. It provides:

- **GPT-124M Architecture**: Standard transformer with RoPE and modern optimizations
- **Tokenized Datasets**: Loading pre-tokenized shards from HuggingFace Hub or local files
- **Modern Optimizers**: Muon optimizer with Newton-Schulz orthogonalization
- **Mixed Precision**: bfloat16 training on A100 for 2x speedup

Training options:
- **Baseline GPT**: Standard residual connections
- **Experimental residual variants**: Optional alternative residual schemes for stability/efficiency

## Quick Reference

| Topic | Reference |
|-------|-----------|
| Model Architecture | [GPT Architecture](references/gpt-architecture.md) |
| Data Loading | [Tokenized Data](references/tokenized-data.md) |
| Optimizers | [Optimizers](references/optimizers.md) |
| Training Loop | [Training Loop](references/training-loop.md) |
| Hyperparameters | [Hyperparameters](references/hyperparameters.md) |

## Installation

```bash
pip install torch einops numpy huggingface_hub
```

## Minimal Example

```python
import modal

app = modal.App("gpt-training")

image = modal.Image.debian_slim(python_version="3.11").pip_install(
    "torch", "einops", "numpy", "huggingface_hub"
)

@app.function(gpu="A100", image=image, timeout=3600)
def train():
    import torch
    from dataclasses import dataclass

    @dataclass
    class GPTConfig:
        block_size: int = 1024
        vocab_size: int = 50257
        n_layer: int = 12
        n_head: int = 12
        n_embd: int = 768
        dropout: float = 0.0
        bias: bool = False

    # Download data, build model, train
    # ... (see references for full implementation)

    return {"final_loss": final_loss}

@app.local_entrypoint()
def main():
    results = train.remote()
    print(results)
```

## Common Imports

The rest: https://github.com/benchflow-ai/skillsbench/blob/HEAD/tasks-extra/mhc-layer-impl/environment/skills/nanogpt-training/SKILL.md