SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.03427 · ACL Findings · 2021

Hierarchical Task Learning from Language Instructions with Unified Transformers and Self-Monitoring

Yichi Zhang, Joyce Chai

arXiv · PDF · Open in the Atlas

Code that ran

4 of 8 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 4 have no contract verdict; 0 failures of our instrument; 4 no recorded run), measured 2026-11-21; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2106.03427.json · badge: /api/ran/2106.03427.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
594zyc/HiTUT canonical (classifier label)
classifier basis: census
4 ran · 4 no recorded run
FunctionStatusWhere it lives
compute_aspect_ratios Ran
ran on a synthesized input
594zyc/HiTUT/models/detector/group_by_aspect_ratio.py
code served (permissive licence) · get_code("e7c827d02d277163")
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, through the batch API) and returned list, 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 torch.utils.data
from PIL import Image
import io

torch.manual_seed(0)

class CustomDataset(torch.utils.data.Dataset):
    def __init__(self, size=2):
        self.size = size
    
    def __len__(self):
        return self.size
    
    def __getitem__(self, idx):
        img = torch.randn(3, 4, 5)
        label = idx
        return img, label
    
    def get_height_and_width(self, idx):
        return 4, 5

class SimpleDataset(torch.utils.data.Dataset):
    def __init__(self, size=2):
        self.size = size
    
    def __len__(self):
        return self.size
    
    def __getitem__(self, idx):
        img = torch.randn(3, 6, 8)
        label = idx
        return img, label

dataset = CustomDataset(size=2)
indices = [0, 1]

ARGS = {
    "dataset": dataset,
    "indices": indices,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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
create_aspect_ratio_groups Ran
ran on a synthesized input
594zyc/HiTUT/models/detector/group_by_aspect_ratio.py
code served (permissive licence) · get_code("b7f47af22eda50af")
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, through the batch API) and returned list, ndim 1, dtype family int; 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
import torch.utils.data

torch.manual_seed(0)

class SimpleDataset(torch.utils.data.Dataset):
    """Minimal dataset that returns (image, label) tuples"""
    def __init__(self, num_samples=5):
        self.num_samples = num_samples
    
    def __len__(self):
        return self.num_samples
    
    def __getitem__(self, idx):
        # Return a simple image tensor (3 channels, 4 height, 6 width) and a label
        img = torch.randn(3, 4, 6)
        label = idx
        return img, label

dataset = SimpleDataset(num_samples=5)

ARGS = {
    "dataset": dataset,
    "k": 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family int — a list 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
get_image_index Ran
ran on a synthesized input
594zyc/HiTUT/models/detector/engine.py
code served (permissive licence) · get_code("72ea0d604062308c")
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, through the batch API) and returned int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 glob
import tempfile
import os

# Create a temporary directory with some mock image files
temp_dir = tempfile.mkdtemp()

# Create a few mock image files matching the pattern '*_image.png'
for i in range(3):
    open(os.path.join(temp_dir, f'file_{i}_image.png'), 'w').close()

# Also create some files that don't match the pattern to ensure glob filters correctly
open(os.path.join(temp_dir, 'other_file.png'), 'w').close()
open(os.path.join(temp_dir, 'file_0_mask.png'), 'w').close()

ARGS = {
    'save_path': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
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
similar Ran
ran on a synthesized input
594zyc/HiTUT/gen/constants.py
code served (permissive licence) · get_code("2f4ceff574612daf")
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, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 'similar' takes two parameters: a and b
# Both are strings that are compared against each other
# The function checks if they belong to the same group in SIMILAR_SET
# or if one is contained in the other (case-insensitive)

ARGS = {
    'a': 'Cup',
    'b': 'Mug'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
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
get_action Not yet run
unverified (harvested, no recorded run)
594zyc/HiTUT/env/reward.py
code served (permissive licence) · get_code("12ba53fedf724ead")
How it ran

no invocation recorded.

get_model_instance_segmentation Not yet run
unverified (harvested, no recorded run)
594zyc/HiTUT/models/detector/mrcnn.py
code served (permissive licence) · get_code("b227dc7436fadba6")
How it ran

no invocation recorded.

get_task Not yet run
unverified (harvested, no recorded run)
594zyc/HiTUT/env/tasks.py
code served (permissive licence) · get_code("5ffac7b83a094577")
How it ran

no invocation recorded.

load_pretrained_model Not yet run
unverified (harvested, no recorded run)
594zyc/HiTUT/models/detector/mrcnn.py
code served (permissive licence) · get_code("e4501c18a317bbf1")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 4 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 4 later) and 4 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.

FunctionAlso present inEarliest-datedOthers named
get_action 5 other papers
2 dated earlier, 3 later
1912.01734 (2019-12)
ALFRED: A Benchmark for Interpreting Grounded Instructions for Everyday Tasks
2105.06453 (2021-05)
Episodic Transformer for Vision-and-Language Navigation
2308.09387 (2023-08)
Multi-Level Compositional Reasoning for Interactive Instruction Following
and 2 more not named here
get_task 3 other papers
1 dated earlier, 2 later
1912.01734 (2019-12)
ALFRED: A Benchmark for Interpreting Grounded Instructions for Everyday Tasks
2308.09387 (2023-08)
Multi-Level Compositional Reasoning for Interactive Instruction Following
2412.11499 (2024-12)
Embodied CoT Distillation From LLM To Off-the-shelf Agents
compute_aspect_ratios 1 other paper
1 later
2203.03890 (2022-03)
ClearPose: Large-scale Transparent Object Dataset and Benchmark
get_model_instance_segmentation 1 other paper
1 later
2412.17288 (2024-12)
Multi-Modal Grounded Planning and Efficient Replanning For Learning Embodied Agents with A Few Examples

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 2026-11-21.

Repositories linked to this paper

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

Abstract

Despite recent progress, learning new tasks through language instructions remains an extremely challenging problem. On the AL-FRED benchmark for task learning, the published state-of-the-art system only achieves a task success rate of less than 10% in an unseen environment, compared to the human performance of over 90%. To address this issue, this paper takes a closer look at task learning. In a departure from a widely applied end-toend architecture, we decomposed task learning into three sub-problems: sub-goal planning, scene navigation, and object manipulation; and developed a model HiTUT 1 (stands for Hierarchical Tasks via Unified Transformers) that addresses each sub-problem in a unified manner to learn a hierarchical task structure. On the ALFRED benchmark, HiTUT has achieved the best performance with a remarkably higher generalization ability. In the unseen environment, HiTUT achieves over 160% performance gain in success rate compared to the previous state of the art. The explicit representation of task structures also enables an in-depth understanding of the nature of the problem and the ability of the agent, which provides insight for future benchmark development and evaluation.

For agents

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

get_harvested_code_for_paper("2106.03427")
get_code_for_paper("2106.03427")
have("2106.03427")

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

curl https://syntology.ai/api/ran/2106.03427.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.03427.svg)](https://syntology.ai/paper/2106.03427)

Connect an agent — have() is free.