Construct ChatGPT Clone with Andrej Karpathy’s nanochat 

What in the event you may construct a practical ChatGPT-like AI for $100? Andrej Karpathy’s new nanochat tells you precisely that! Launched on October 13, 2025, Karpathy’s nanochat challenge is an open-source LLM coded in roughly 8,000 traces of PyTorch. It provides you an easy roadmap on find out how to practice a language mannequin from scratch and make your individual personal AI in a few hours. On this article, we’ll discuss concerning the newly launched nanochat and find out how to correctly set it up for the coaching step-by-step. 

What’s nanochat?

The nanochat repository supplies a full-stack pipeline to coach a minimal ChatGPT clone. It takes care of every little thing from tokenization to the tip net consumer interface. This technique is a successor to the earlier nanoGPT. It introduces key options similar to supervised fine-tuning (SFT), reinforcement studying (RL), and enhanced inference. 

Key Options

The challenge has quite a few important elements. It incorporates a brand new Rust-built tokenizer for top efficiency. The coaching pipeline employs high quality knowledge similar to FineWeb-EDU for pretraining. It additionally employs specialised knowledge similar to SmolTalk and GSM8K for post-training fine-tuning. For safety, the mannequin can run code inside a Python sandbox. 

The challenge works properly inside your price range. The basic “speedrun” mannequin is round $100 and trains for 4 hours. You can even develop a extra sturdy mannequin for about $1,000 with roughly 42 hours of coaching. 

Efficiency

The efficiency will increase with the coaching time. 

• 4 hours: The fast run provides you a easy conversational mannequin. It might compose easy poems or describe ideas similar to Rayleigh scattering. 

A number of the abstract metrics had been produced by the $100 speedrun for 4 hours. 

• 12 hours: The mannequin begins to surpass GPT-2 on the CORE benchmark. 
• 24 hours: It will get respectable scores, similar to 40% on MMLU and 70% on ARC-Simple. 

The first instructional intention of the nanochat challenge is to offer a simple, hackable baseline. This makes it an excellent useful resource for college kids, researchers, and AI hobbyists. 

Conditions and Setup

Earlier than you begin, you need to prepared your {hardware} and software program. It’s simple to do with the right instruments. 

{Hardware} Necessities

The challenge is finest dealt with by an 8xH100 GPU node. These can be found on suppliers similar to Lambda GPU Cloud for about $24 an hour. You can even use a single GPU with gradient accumulation. This can be a slower technique, however eight occasions slower. 

Software program

You’ll require a typical Python setting together with PyTorch. The challenge depends upon the uv bundle supervisor to handle dependencies. Additionally, you will require Git put in with the intention to clone the repository. As an non-compulsory alternative, you could embody Weights & Biases for logging your coaching runs. 

Preliminary Steps

Cloning the official repository comes first:  

git clone [email protected]:karpathy/nanochat.git 

Second, become the challenge listing, i.e, nanochat, and set up the dependencies. 

cd nanochat 

Lastly, create and fix to your cloud GPU occasion to begin coaching. 

Information for Coaching Your Personal ChatGPT Clone

What follows is a step-by-step information to coaching your very first mannequin. Paying shut consideration to those steps will yield a working LLM. The official walkthrough within the repository incorporates extra info. 

Step 1: Atmosphere Preparation

First, boot your 8xH100 node. As soon as up, set up uv bundle supervisor utilizing the provided script. It’s good to have long-running issues inside a display session. This makes the coaching proceed even whenever you disconnect. 

# set up uv (if not already put in) 
command -v uv &> /dev/null || curl -LsSf https://astral.sh/uv/set up.sh | sh 
# create a .venv native digital setting (if it would not exist) 
[ -d ".venv" ] || uv venv 
# set up the repo dependencies 
uv sync 
# activate venv in order that `python` makes use of the challenge's venv as a substitute of system python 
supply .venv/bin/activate 

Step 2: Knowledge and Tokenizer Setup

First, we have to set up Rust/Cargo in order that we are able to compile our customized Rust tokenizer. 

# Set up Rust / Cargo 
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y 
supply "$HOME/.cargo/env" 
# Construct the rustbpe Tokenizer 
uv run maturin develop --release --manifest-path rustbpe/Cargo.toml 

The pretraining knowledge is simply the textual content of a number of webpages, and for this half, we’ll use the FineWeb-EDU dataset. However Karpathy recommends utilizing the next model. 

https://huggingface.co/datasets/karpathy/fineweb-edu-100b-shuffle

python -m nanochat.dataset -n 240 

As soon as downloaded, you practice the Rust tokenizer on a big corpus of textual content. This step is made to be quick by the script. It ought to compress to roughly a 4.8 to 1 compression ratio. 

python -m scripts.tok_train --max_chars=2000000000 
python -m scripts.tok_eval 

Step 3: Pretraining

Now, you need to obtain the analysis knowledge bundle. That is the place the take a look at datasets for the mannequin’s efficiency reside. 

curl -L -o eval_bundle.zip https://karpathy-public.s3.us-west-2.amazonaws.com/eval_bundle.zip 
unzip -q eval_bundle.zip 
rm eval_bundle.zip 
mv eval_bundle "$HOME/.cache/nanochat" 

Additionally, setup wandb for seeing good plots throughout coaching. uv already put in wandb for us up above, however you continue to should arrange an account and log in with: 

wandb login 

Now you could provoke the principle pretraining script. Execute it with the torchrun command to leverage all eight GPUs. The method trains the mannequin on easy language patterns from the FineWeb-EDU corpus. This stage requires round two to a few hours for speedrun. This can be a very important a part of the method for coaching a language mannequin. 

torchrun --standalone --nproc_per_node=8 -m scripts.base_train -- --depth=20 

We’re initiating coaching on 8 GPUs utilizing the scripts/base_train.py script. The mannequin is a 20-layer Transformer. Every GPU handles 32 sequences of 2048 tokens per ahead and backward go, giving a complete of 32 × 2048 = 524,288 (≈0.5M) tokens processed per optimization step. 

If Weights & Biases (wandb) is configured, you possibly can add the –run=speedrun flag to assign a run identify and allow logging. 

When coaching begins, you’ll see an output just like the next (simplified right here for readability): 

Step 4: Midtraining and SFT

As soon as pretraining, you proceed to midtraining. Midtraining applies the SmolTalk dataset to offer the mannequin with extra conversational energy. After that, you’ll conduct supervised fine-tuning (SFT) on knowledge similar to GSM8K. That is what aids the mannequin in studying to execute directions in addition to fixing issues. 

We are able to begin the mid-training as follows: this run solely takes about 8 minutes, so much shorter than pre-training at ~3 hours. 

torchrun --standalone --nproc_per_node=8 -m scripts.mid_train 

After mid-training comes the Finetuning stage. This section entails one other spherical of finetuning on conversational knowledge, however with a concentrate on deciding on solely the highest-quality, most well-curated examples. It’s additionally the stage the place safety-oriented changes are made, similar to coaching the mannequin on applicable refusal behaviors for delicate or restricted queries. This once more solely runs for about 7 minutes. 

torchrun --standalone --nproc_per_node=8 -m scripts.chat_sft 

Step 5: Non-obligatory RL 

The nanochat open-source LLM additionally has preliminary reinforcement studying help. You’ll be able to run a way referred to as GRPO on the GSM8K dataset. That is an non-compulsory course of and should take one other hour. Test that Karpathy stated RL help remains to be in its infancy. 

torchrun --standalone --nproc_per_node=8 -m scripts.chat_rl 

Step 6: Inference and UI

With coaching completed, now you can run the inference script. This allows you to discuss to your mannequin utilizing an online UI or command-line interface. Attempt operating it with some examples like “Why is the sky blue?” to expertise your creation. 

python -m scripts.chat_cli  (for Command line window) 

OR  

python -m scripts.chat_web. (for Internet UI) 

The chat_web script will serve the Engine utilizing FastAPI. Ensure that to entry it appropriately, e.g., on Lambda, use the general public IP of the node you’re on, adopted by the port, so for instance http://209.20.xxx.xxx:8000/, and so forth. 

Step 7: Overview Outcomes

Now, testing it with the net interface on the hyperlink on which the nanochat is hosted. 

Lastly, have a look at the report.md within the repository. It has some necessary metrics to your mannequin, similar to its CORE rating and GSM8K accuracy. The bottom speedrun runs for about $92.40 to place in a bit lower than 4 hours of labor. 

Word: I’ve taken the code and steps from Andrej Karapathy’s nano chat GitHub. You’ll find full documentation right here. What I showcased above is an easier and shorter model.

Customizing and Scaling

The speedrun is a superb start line. From that time, you possibly can additional customise the mannequin. This is likely one of the most vital benefits of Karpathy’s nanochat launch. 

Tuning Choices

You’ll be able to tweak the depth of the mannequin to enhance efficiency. With the --depth=26 flag, say, you step right into a extra highly effective $300 vary. You may additionally attempt utilizing different datasets or altering coaching hyperparameters. 

Scaling Up

The repository particulars a $1,000 stage. This entails an prolonged coaching run of roughly 41.6 hours. It yields a mannequin with improved coherence and better benchmark scores. In case you are dealing with VRAM constraints, try to decrease the --device_batch_size setting. 

Personalization Challenges

Others can fine-tune the mannequin on private knowledge. Karpathy advises towards this, as this may find yourself producing “slop.” A greater approach to make use of private knowledge is retrieval-augmented technology (RAG) by way of instruments similar to NotebookLM. 

Conclusion

The nanochat challenge allows each researchers and rookies. It provides an affordable and easy method to practice a robust open-source LLM. With a restricted price range and an open weekend, you possibly can go from setup to deployment. Use this tutorial to coach your individual ChatGPT, try the nanochat repository, and take part in the neighborhood discussion board to assist out. Your journey to coach a language mannequin begins right here. 

Regularly Requested Questions

Harsh Mishra is an AI/ML Engineer who spends extra time speaking to Massive Language Fashions than precise people. Obsessed with GenAI, NLP, and making machines smarter (so that they don’t exchange him simply but). When not optimizing fashions, he’s in all probability optimizing his espresso consumption. 🚀☕