SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2309.04459 · NeurIPS · 2024

Subwords as Skills: Tokenization for Sparse-Reward Reinforcement Learning

Matthew Walter, David Yunis, Justin Jung, Z Dai

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 failures of our instrument; 0 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2309.04459.json · badge: /api/ran/2309.04459.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
dyunis/subwords_as_skills canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
huggingface_tokenize_trajectories Ran
ran; our contract draft was wrong, not the code
dyunis/subwords_as_skills/subwords.py
code served (permissive licence) · get_code("18e68313c9af3043")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from tokenizers import models, normalizers, pre_tokenizers, processors, trainers, Tokenizer

torch.manual_seed(0)

# Create small trajectories as lists of integers (which will be converted to characters)
# Each trajectory is a sequence of small integers that can be converted to chr()
trajectories = [
    [65, 66, 67, 68, 69],  # "ABCDE"
    [70, 71, 72, 73, 74],  # "FGHIJ"
    [75, 76, 77, 78, 79],  # "KLMNO"
    [80, 81, 82, 83, 84],  # "PQRST"
]

vocab_size = 50

mode = 'bpe'

ARGS = {
    'trajectories': trajectories,
    'vocab_size': vocab_size,
    'mode': mode,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, tokenizers, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
random_k_actions_like Ran
ran; our contract draft was wrong, not the code
dyunis/subwords_as_skills/subwords.py
code served (permissive licence) · get_code("c50cb75f323104e9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned list. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Analyze the function to understand what vocab should be:
# - vocab is a dict where values() are iterables of numbers
# - It's used to extract primitives (unique numbers across all values)
# - It's used to find compounds (values with len > 1)
# - Each compound sequence is used with len(seq) to determine random choice size

# Create a minimal vocab dict
vocab = {
    'action_1': [1, 2],           # compound: len=2
    'action_2': [2, 3, 4],        # compound: len=3
    'primitive_1': [1],           # primitive: len=1
    'primitive_2': [2],           # primitive: len=1
    'primitive_3': [3],           # primitive: len=1
    'primitive_4': [4],           # primitive: len=1
}

ARGS = {
    'vocab': vocab,
    'seed': 0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list — a list output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
to_str Ran
ran; our contract draft was wrong, not the code
dyunis/subwords_as_skills/subwords.py
code served (permissive licence) · get_code("e0e45cd4b9d51aa0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function to_str takes an int_list and converts each integer to its
# corresponding character using chr(), then joins them into a string.
# We need to provide a list of integers that are valid Unicode code points.

int_list = [72, 101, 108, 108, 111]  # "Hello"

ARGS = {
    "int_list": int_list
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Exploration in sparse-reward reinforcement learning (RL) is difficult due to the need for long, coordinated sequences of actions in order to achieve any reward. Skill learning, from demonstrations or interaction, is a promising approach to address this, but skill extraction and inference are expensive for current methods. We present a novel method to extract skills from demonstrations for use in sparsereward RL, inspired by the popular Byte-Pair Encoding (BPE) algorithm in natural language processing. With these skills, we show strong performance in a variety of tasks, 1000× acceleration for skill-extraction and 100× acceleration for policy inference. Given the simplicity of our method, skills extracted from 1% of the demonstrations in one task can be transferred to a new loosely related task. We also note that such a method yields a finite set of interpretable behaviors. Our code is available at https://github.com/dyunis/subwords_as_skills. (a) AntMaze (b) Kitchen Figure 1: A sample of some "skills" that our method identifies for the (a) AntMaze and (b) Kitchen environments, where color is darker for poses earlier in the trajectory. Skills consist of linear motion and turning in AntMaze, and reaching and pulling motions in Kitchen. Our method discovers a finite inventory of skills, so it is possible to visualize and interpret them.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2309.04459")
get_code_for_paper("2309.04459")
have("2309.04459")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2309.04459.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2309.04459.svg)](https://syntology.ai/paper/2309.04459)

Connect an agent — have() is free.