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.

Safety Notice

This listing is from the official public ClawHub registry. Review SKILL.md and referenced scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "nanogpt-training" with this command: npx skills add lnj22/mhc-layer-impl-nanogpt-training

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

TopicReference
Model ArchitectureGPT Architecture
Data LoadingTokenized Data
OptimizersOptimizers
Training LoopTraining Loop
HyperparametersHyperparameters

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

ScenarioApproach
Standard GPT trainingUse baseline model with standard residuals
Stability experimentsTry alternative residual variants or extra streams
Small experimentsUse T4/A10G GPU
Full trainingUse A100 with bfloat16
Custom dataModify the dataset loader class
Different model sizeAdjust GPTConfig parameters

Metrics to Monitor

MetricTypical SignalNotes
Validation lossSteady decreaseAbsolute value depends on dataset/tokenizer
Grad normModerate, stable rangeLarge spikes indicate instability
Training stabilitySmooth curvesFrequent spikes suggest LR/batch issues
ThroughputConsistent tokens/secUse for comparing configs

External Resources

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

Huo15 Openclaw Enhance

火一五·克劳德·龙虾增强插件 v5.7.8 — 全面适配 openclaw 2026.4.24:peerDep ^4.24 + build/compat 同步到 4.24 + 14 处 api.on 全部去掉 as any 改成 typed hook(hookName 联合类型 + handler 自动推断 Pl...

Registry SourceRecently Updated
General

Content Trend Analyzer

Aggregates and analyzes content trends across platforms to identify hot topics, user intent, content gaps, and generates data-driven article outlines.

Registry SourceRecently Updated
General

Prompt Debugger

Debug prompts that produce unexpected AI outputs — diagnose failure modes, identify ambiguity and conflicting instructions, test variations, compare model re...

Registry SourceRecently Updated
General

Indie Maker News

独行者 Daily - 变现雷达。读对一条新闻,少走一年弯路。每天5分钟,给创业者装上商业雷达。聚焦一人公司、副业、创业变现资讯,智能分类,行动导向。用户下载即能用,无需本地部署!

Registry SourceRecently Updated