Agent skill
nanogpt-training
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.
Install this agent skill to your Project
npx add-skill https://github.com/benchflow-ai/skillsbench/tree/main/tasks-no-skills/mhc-layer-impl/environment/skills/nanogpt-training
SKILL.md
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 |
| Data Loading | Tokenized Data |
| Optimizers | Optimizers |
| Training Loop | Training Loop |
| Hyperparameters | Hyperparameters |
Installation
pip install torch einops numpy huggingface_hub
Minimal Example
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
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.cuda.amp import autocast, GradScaler
from dataclasses import dataclass
from einops import rearrange, repeat, reduce
import numpy as np
import math
When to Use What
| Scenario | Approach |
|---|---|
| Standard GPT training | Use baseline model with standard residuals |
| Stability experiments | Try alternative residual variants or extra streams |
| Small experiments | Use T4/A10G GPU |
| Full training | Use A100 with bfloat16 |
| Custom data | Modify the dataset loader class |
| Different model size | Adjust GPTConfig parameters |
Metrics to Monitor
| Metric | Typical Signal | Notes |
|---|---|---|
| Validation loss | Steady decrease | Absolute value depends on dataset/tokenizer |
| Grad norm | Moderate, stable range | Large spikes indicate instability |
| Training stability | Smooth curves | Frequent spikes suggest LR/batch issues |
| Throughput | Consistent tokens/sec | Use for comparing configs |
External Resources
- nanoGPT: https://github.com/karpathy/nanoGPT
- build-nanogpt: https://github.com/karpathy/build-nanogpt
- modded-nanogpt: https://github.com/KellerJordan/modded-nanogpt
- FineWeb-Edu token shards: https://huggingface.co/datasets/karpathy/fineweb-edu-100B-gpt2-token-shards
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
benchflow-ai/skillsbench
csv-processing
Use this skill when reading sensor data from CSV files, writing simulation results to CSV, processing time-series data with pandas, or handling missing values in datasets.
benchflow-ai/skillsbench
pid-controller
Use this skill when implementing PID control loops for adaptive cruise control, vehicle speed regulation, throttle/brake management, or any feedback control system requiring proportional-integral-derivative control.
benchflow-ai/skillsbench
yaml-config
Use this skill when reading or writing YAML configuration files, loading vehicle parameters, or handling config file parsing with proper error handling.
benchflow-ai/skillsbench
simulation-metrics
Use this skill when calculating control system performance metrics such as rise time, overshoot percentage, steady-state error, or settling time for evaluating simulation results.
benchflow-ai/skillsbench
vehicle-dynamics
Use this skill when simulating vehicle motion, calculating safe following distances, time-to-collision, speed/position updates, or implementing vehicle state machines for cruise control modes.
benchflow-ai/skillsbench
web-interface-guidelines
Vercel's comprehensive UI guidelines for building accessible, performant web interfaces. Use this skill when reviewing or building UI components for compliance with best practices around accessibility, performance, animations, and visual stability.
Didn't find tool you were looking for?