nanochat-llm-training

# Run the reference pipeline: data download, pretraining, SFT, eval, chat
bash runs/speedrun.sh

OMP_NUM_THREADS=1 torchrun --standalone --nproc_per_node=8 -m scripts.base_train -- \
    --depth=26 \
    --run="d26_run" \
    --model-tag="d26"

python -m scripts.base_train -- \
    --depth=26 \
    --run="d26_single"

OMP_NUM_THREADS=1 torchrun --standalone --nproc_per_node=8 -m scripts.base_train -- \
    --depth=12 \
    --run="d12_exp" \
    --model-tag="d12" \
    --core-metric-every=999999 \
    --sample-every=-1 \
    --save-every=-1

bash runs/runcpu.sh

# After training completes
source .venv/bin/activate
python -m scripts.chat_web
# Visit http://<your-server-ip>:8000/

python -m scripts.chat_cli -p "hello"

bash runs/scaling_laws.sh   # sweep depths for scaling law data
bash runs/miniseries.sh     # train full compute-optimal miniseries

# Smaller/faster experiments
python -m scripts.base_train -- --depth=12 --run="quick_test"
 
# Full GPT-2 grade
torchrun --standalone --nproc_per_node=8 -m scripts.base_train -- --depth=26 --run="gpt2_repro"

# Force fp32 for inference
NANOCHAT_DTYPE=float32 python -m scripts.chat_cli -p "hello"
 
# Force bf16 for training
NANOCHAT_DTYPE=bfloat16 torchrun --nproc_per_node=8 -m scripts.base_train
 
# float16 training (enables GradScaler automatically)
NANOCHAT_DTYPE=float16 torchrun --nproc_per_node=8 -m scripts.base_train

nanochat/
├── gpt.py              # GPT nn.Module Transformer
├── engine.py           # Inference with KV Cache
├── dataloader.py       # Tokenizing Distributed Data Loader
├── dataset.py          # Download/read utils for pretraining data
├── optim.py            # AdamW + Muon optimizer (1GPU and distributed)
├── core_eval.py        # DCLM CORE score evaluation
├── loss_eval.py        # Bits-per-byte evaluation
├── checkpoint_manager.py  # Save/Load checkpoints
├── common.py           # Utilities, COMPUTE_DTYPE
├── execution.py        # Python code execution tool for LLM
└── engine.py           # Efficient KV-cache inference

scripts/
├── base_train.py       # Pretraining entry point
├── chat_web.py         # Web chat UI server
└── chat_cli.py         # CLI chat interface

runs/
├── speedrun.sh         # Reference full pipeline (GPT-2 speedrun)
├── scaling_laws.sh     # Scaling law sweeps
├── miniseries.sh       # Full compute-optimal miniseries
└── runcpu.sh           # CPU/MPS example

import torch
from nanochat.gpt import GPT
from nanochat.engine import InferenceEngine
from nanochat.checkpoint_manager import CheckpointManager
 
# Load checkpoint
ckpt_manager = CheckpointManager("checkpoints/d26")
model, config = ckpt_manager.load()
model.eval()
 
# Run inference with KV cache
engine = InferenceEngine(model)
output = engine.generate(
    prompt="Once upon a time",
    max_new_tokens=200,
    temperature=0.8,
    top_p=0.95,
)
print(output)

import subprocess
 
def train_model(depth: int, run_name: str, nproc: int = 8):
    """Launch a compute-optimal training run for given depth."""
    cmd = [
        "torchrun",
        "--standalone",
        f"--nproc_per_node={nproc}",
        "-m", "scripts.base_train",
        "--",
        f"--depth={depth}",
        f"--run={run_name}",
        f"--model-tag={run_name}",
    ]
    subprocess.run(cmd, env={"OMP_NUM_THREADS": "1", **__import__("os").environ})
 
# Quick research iteration
train_model(depth=12, run_name="my_experiment_d12")
 
# Full GPT-2 grade
train_model(depth=26, run_name="my_gpt2_repro")

# Default device_batch_size=32 needs ~80GB VRAM per GPU
# Reduce for smaller GPUs (gradient accumulation handles the rest)
torchrun --standalone --nproc_per_node=4 -m scripts.base_train -- \
    --depth=12 \
    --device_batch_size=16 \
    --run="low_vram_run"
 
# Even smaller
python -m scripts.base_train -- \
    --depth=8 \
    --device_batch_size=4 \
    --run="single_gpu_small"

# nanochat logs to wandb automatically. Key metrics to watch:
# - val_bpb: validation loss in bits-per-byte (vocab-size-invariant)
#   as a function of step, total_training_time, total_training_flops
# - core_metric: DCLM CORE score (target > 0.2565 to beat GPT-2)
# - train/mfu: Model FLOPS utilization
# - train/tok_per_sec: Training throughput
 
# Set wandb project via env var before training
import os
os.environ["WANDB_PROJECT"] = "my-nanochat-runs"

# dev/gen_synthetic_data.py — generate identity/personality data
# Then mix into SFT stage per the guide:
# https://github.com/karpathy/nanochat/discussions/139
 
# Example: generate data and point SFT to it
python dev/gen_synthetic_data.py --output data/identity_sft.jsonl
# Then reference in your SFT script configuration

# 1. Make a code change in nanochat/
# 2. Run quick d12 to validate
OMP_NUM_THREADS=1 torchrun --standalone --nproc_per_node=8 -m scripts.base_train -- \
    --depth=12 --run="test_my_change" \
    --core-metric-every=999999 --sample-every=-1 --save-every=-1
# 3. Check wandb: val_bpb vs step/time/flops
# 4. If promising, test at d16 or d26

# FP8 is used in the speedrun for additional speedup
# See runs/speedrun.sh for the exact invocation
bash runs/speedrun.sh

python -m nanochat.core_eval --checkpoint checkpoints/d26/latest

# On remote machine after training:
source .venv/bin/activate
python -m scripts.chat_web
# Access via: http://<PUBLIC_IP>:8000/
# Use `screen` or `tmux` to keep alive
screen -S nanochat
python -m scripts.chat_web
# Ctrl+A, D to detach

# Reduce --device_batch_size (default 32)
# Code uses gradient accumulation to maintain effective batch size
--device_batch_size=16   # Try 16, 8, 4, 2, 1

# MPS (Apple Silicon) or CPU — use runcpu.sh as template
bash runs/runcpu.sh
# Results will be weak; this is for development/debugging only

# nanochat auto-enables GradScaler when NANOCHAT_DTYPE=float16
NANOCHAT_DTYPE=float16 torchrun --nproc_per_node=8 -m scripts.base_train -- --depth=12
# Note: RL scripts do NOT support float16 (SFT and base_train do)

# Default falls back to float32; optionally use float16
NANOCHAT_DTYPE=float16 torchrun --nproc_per_node=8 -m scripts.base_train -- --depth=12

# Ensure the port (default 8000) is open in your cloud provider's firewall/security group
# Use the public IP, not localhost:
# http://<PUBLIC_IP>:8000/