gpt1900-d26-22btok

GPT-1900 (D26, 22B tokens)

A 1.2B parameter GPT-style language model trained on ~22B tokens of pre-1900 English text (20x data:param ratio).

Model Details

• Architecture: Custom GPT with RoPE, QK-norm, ReLU², value embeddings (ResFormer), per-layer residual/skip scalars
• Parameters: ~1.2B
• Layers: 26
• Hidden dim: 1664
• Attention heads: 13 (query) / 13 (kv)
• Head dim: 128
• Context length: 2048 tokens
• Vocab size: 32,768 (BPE, GPT-4 style split pattern)
• Training: FP8 (tensorwise), Muon+AdamW optimizer
• Final val BPB: 0.802

Checkpoint Contents

model_017517.pt          # Model weights
meta_017517.json         # Training config and metadata
optim_017517_rank*.pt    # Optimizer state shards (for resuming training)
tokenizer/               # BPE tokenizer (tiktoken format) + token byte counts
nanochat/                # Source code to load and run the model

Quick Start

import torch
from nanochat.gpt import GPT, GPTConfig
from nanochat.tokenizer import RustBPETokenizer

# Load tokenizer
tokenizer = RustBPETokenizer.from_directory("tokenizer")

# Load model
import json
with open("meta_017517.json") as f:
    meta = json.load(f)

config = GPTConfig(**meta["model_config"])

with torch.device("meta"):
    model = GPT(config)
model.to_empty(device="cuda")
model.init_weights()

state_dict = torch.load("model_017517.pt", map_location="cuda")
state_dict = {k.removeprefix("_orig_mod."): v for k, v in state_dict.items()}
model.load_state_dict(state_dict, strict=True, assign=True)
model.eval()

# Generate
bos = tokenizer.get_bos_token_id()
tokens = tokenizer.encode("It was a dark and stormy night", prepend=bos)
with torch.amp.autocast(device_type="cuda", dtype=torch.bfloat16):
    for token in model.generate(tokens, max_tokens=100, temperature=0.8):
        print(tokenizer.decode([token]), end="", flush=True)

Dependencies

torch>=2.9
tiktoken
rustbpe

Training

Trained with the nanochat framework on H100 GPUs. To resume training, load the optimizer shards (optim_017517_rank*.pt) — one per rank.