Reporium
GraphWikiTaxonomyStacksInsightsTrendsArchitectureAI-NativeFAQ
Ask anything about the repo library…
Loading repo…
←Library/nanochat
Library/nanochatForked

karpathy/nanochat

nanochat

The best ChatGPT that $100 can buy.

View on GitHub↗Upstream karpathy/nanochat↗

Builder

karpathy

karpathy

karpathy • individual

Stars

54,381

Using upstream star count

Forks

7,359

Using upstream fork count

Open Issues

0

Activity Score

0/100

0 commits in 30d

Created

Oct 13, 2025

Project creation date

README Summary

nanochat is the simplest experimental harness for training LLMs. It is designed to run on a single GPU node, the code is minimal/hackable, and it covers all major LLM stages including tokenization, pretraining, finetuning, evaluation, inference, and a chat UI. For example, you can train your own GPT-2 capability LLM (which cost ~$43,000 to train in 2019) for only $48 (~2 hours of 8XH100 GPU node) and then talk to it in a familiar ChatGPT-like web UI. On a spot instance, the total cost can be clo

Community Evaluation

Loading…

AI Dev Skills

Unmapped

Conversational AICost-Optimized AI DeploymentLanguage Model ImplementationNatural Language ProcessingTransformer Architecture

Tags

Conversational AICost-Optimized AI DeploymentLanguage Model ImplementationNatural Language ProcessingTransformer ArchitectureBackendBenchmarkingEmbeddingsEvalsFine-TuningForkedGPTGPU / CUDAHuggingFaceHumanEvalKV CacheLarge Language ModelsMMLUOpenAIPyTorchPythonReinforcement LearningSynthetic DataTransformersTutorialWeights & Biases

Taxonomy

AI Trends

Small Language ModelsCost-Efficient AIAccessible AI Development

category

Foundation ModelsRAG & RetrievalModel TrainingEvals & BenchmarkingObservability & MonitoringInference & ServingDev Tools & AutomationLearning Resources

Deployment Context

Self-hosted

Modalities

Text

Skill Areas

Conversational AILanguage Model ImplementationNatural Language ProcessingTransformer ArchitectureCost-Optimized AI Deployment

tag

BackendBenchmarkingEmbeddingsEvalsFine-TuningForkedGPTGPU / CUDAHuggingFaceHumanEvalKV CacheLarge Language ModelsMMLUOpenAIPyTorchPythonReinforcement LearningSynthetic DataTransformersTutorialWeights & Biases

Use Cases

Budget-Constrained Chatbot DevelopmentEducational AI ImplementationPersonal Assistant CreationCost-Effective Customer Service Automation

Recent Activity

Updated 2 months ago

7 Days

0

30 Days

0

90 Days

8

submit new time to GPT-2 leaderboard entry: 99 minutes

Andrej Karpathy • Mar 14, 2026

1b1cc3c

Autoresearch round 2: smear, backout, and hyperparameter tuning

Andrej Karpathy • Mar 14, 2026

a825e63

new leaderboard entry coming from improvements of autoresearch round 1, time to gpt-2 from 2.02 hour

Andrej Karpathy • Mar 10, 2026

f068604

Quality

prototype
Quality
low
Maturity
prototype

Categories

RAG & RetrievalPrimaryEvals & BenchmarkingObservability & MonitoringInference & ServingDev Tools & AutomationLearning ResourcesFoundation ModelsModel TrainingSafety & AlignmentOther AI / ML

PM Skills

Cost & EfficiencyData & EvaluationProduct Discovery

Languages

Python100.0%

Timeline

Project created
Oct 13, 2025
Forked
Mar 14, 2026
Your last push
2 months ago
Upstream last push
29 days ago
Tracked since
Mar 17, 2026

Similar Repos

pgvector cosine similarity · $0

Loading…