dyck-k128-seq_len_2048-1B

Procedurally generated k-shuffle Dyck bracket sequences (Hu et al. 2025, arXiv:2502.19249), as flat uint16 token-id .bin files. Token ids are 0-based: opening bracket type i is id i and its matching close is i + k, so ids span [0, 2k) and the vocabulary is 2k = 256.

Grammar parameters

param	value
k (bracket types)	128
max_depth	16
p_open	0.5
seq_length	2048

file	split	tokens
train.bin	train	999,999,488
val.bin	val	10,000,384

The bin is a concatenation of fixed-length 2048-token words, each a complete depth-0-starting Dyck word (so num_tokens is a whole multiple of seq_length). Read it in seq_length-token chunks — tokens.reshape(-1, 2048) — to train one word per row, exactly as generated. train (seed 0) and val (a disjoint generator seed) are independent streams of the same grammar. Token count = filesize / 2; train.meta.json / val.meta.json carry the full config.

Load a bin with the standard Hugging Face downloader:

from huggingface_hub import hf_hub_download
import numpy as np

path = hf_hub_download(repo_id="stanpony/dyck-k128-seq_len_2048-1B", filename="train.bin", repo_type="dataset")
tokens = np.memmap(path, dtype="uint16", mode="r")