Nanochat Llm Training
aradotso/trending-skillsYou can train your own small language model for less than one hundred dollars. Nanochat is a simple tool that handles everything from start to finish. It was created by Andrej Karpathy and runs on a single powerful computer.
The tool covers tokenization, training, fine tuning, evaluation, and a chat interface. You can adjust just one setting called depth to control the model size. This makes it easy to experiment with different scales.
Nanochat lets you reproduce the capability of GPT-2 from 2019. That original model cost about forty three thousand dollars to train. With nanochat you can do it for roughly forty eight dollars on modern hardware.
Add Nanochat Llm Training skill to your workflow
Global
mkdir -p ~/.claude/skills/nanochat-llm-trainingProject
mkdir -p .claude/skills/nanochat-llm-trainingSource Repository
More Data Science & ML Claude Code Skills
Explore Runlllllllama/rigorpilot-skills
Plan and rank quick deep learning exploration runs with clear boundaries
Run Trainlllllllama/rigorpilot-skills
Run deep learning training conservatively with structured logs and metrics
Env And Assets Bootstraplllllllama/rigorpilot-skills
Safe and careful setup of conda environments and asset paths for deep learning reproduction
Env And Assets Bootstraplllllllama/ai-paper-reproduction-skill
Prepare conda-first environment and asset paths for deep learning reproduction
Data Visualizationanthropics/knowledge-work-plugins
Pick the right chart, write clean Python code, and design for everyone
Powerbi Modelinggithub/awesome-copilot
Expert guidance for building optimized Power BI semantic models with star schemas and DAX
Power Bi Dax Optimizationgithub/awesome-copilot
Make your Power BI DAX formulas faster and easier to maintain
Bigquery Basicsgoogle/skills
Easily analyze massive datasets with SQL and built-in machine learning