SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2001.00248 · ICLR · 2020

Meta Reinforcement Learning with Autonomous Inference of Subtask Dependencies

Sungryull Sohn, Hyunjae Woo, Jongwook Choi, Honglak Lee

arXiv · PDF · Open in the Atlas

Code that ran

5 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 2 failures of our instrument; 0 no recorded run), measured 2026-11-22; 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/2001.00248.json · badge: /api/ran/2001.00248.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
srsohn/msgi canonical (classifier label)
classifier basis: classifier_2025
5 ran · 0 no recorded run
FunctionStatusWhere it lives
ILP Ran
ran on a synthesized input · constructed an object; no result was computed
srsohn/msgi/grid-world/graph/ILP.py
code served (permissive licence) · get_code("e1cd54d98edcfe1f")
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 an instance of its own class, ILP: the run constructed the object and did not call it, so no result was computed. 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
import numpy as np

torch.manual_seed(0)

class MockEnv:
    def __init__(self):
        self.ntasks = 3
    
    def get_tid_to_tind(self):
        return {0: 0, 1: 1, 2: 2}
    
    def get_subtask_lists(self):
        return [[0], [1], [2]]

ARGS = {
    "args": type('Args', (), {
        'num_processes': 2,
        'act_dim': 4,
        'tr_epi': 5,
        'bonus_mode': 1
    })()
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
ipdb, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ILP — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ILP 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
_to_multi_hot Ran
ran; our contract draft was wrong, not the code
srsohn/msgi/grid-world/graph/ILP.py
code served (permissive licence) · get_code("36db2ebee9fad758")
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 Tensor, ndim 2, dtype family float, shape (4, 4) 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
import numpy as np

torch.manual_seed(0)

# Test case 1: 1D numpy array input
index_tensor_1d_np = np.array([0, 2, 1, 3])
max_dim_1 = 4

# Test case 2: 1D torch tensor input
index_tensor_1d_torch = torch.tensor([1, 2, 0])
max_dim_2 = 4

# Test case 3: 2D torch tensor input
index_tensor_2d_torch = torch.tensor([[0, 1], [2, 3]])
max_dim_3 = 4

# Using the first test case as the primary example
ARGS = {
    "index_tensor": index_tensor_1d_np,
    "max_dim": max_dim_1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (4, 4) — shape (4, 4) 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
_transform Ran
ran; our contract draft was wrong, not the code
srsohn/msgi/grid-world/graph/ILP.py
code served (permissive licence) · get_code("81e43e94407ed380")
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 Tensor, ndim 2, dtype family float, shape (4, 1) 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)

# Create input_tensor: can be 1D or 2D
# The function handles 1D by unsqueezing to 2D
# Let's test with 1D input
input_tensor = torch.tensor([0, 1, 2, 1])  # shape: (4,)

# Create mapping tensor
# gather(1, input_tensor) gathers along dimension 1
# mapping should have shape (batch_size, vocab_size, ...)
# input_tensor after unsqueeze will be (4, 1)
# So mapping needs to support indexing where dim 1 has at least max(input_tensor)+1 elements
# Let's make mapping shape (4, 5) - batch of 4, with 5 possible values to gather
mapping = torch.randn(4, 5)

ARGS = {
    "input_tensor": input_tensor,
    "mapping": mapping,
}
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 Tensor, ndim 2, dtype family float, shape (4, 1) — shape (4, 1) 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
batch_bin_encode Ran
ran, honoured the contract we drafted
srsohn/msgi/grid-world/graph/ILP.py
code served (permissive licence) · get_code("9fd224ed549d872d")
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 Tensor, ndim 1, dtype family int, shape (2,) 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
import numpy as np

torch.manual_seed(0)

# Test case 1: 2D torch tensor (batch of binary vectors)
bin_tensor_2d = torch.tensor([
    [1, 0, 1, 1],
    [0, 1, 0, 1]
], dtype=torch.float32)

# Test case 2: 1D torch tensor (single binary vector)
bin_tensor_1d = torch.tensor([1, 0, 1, 0], dtype=torch.float32)

# Test case 3: numpy 1D array (single binary vector)
bin_tensor_np = np.array([1, 1, 0, 1], dtype=np.int32)

# Use the 2D torch tensor as the main test case
ARGS = {
    'bin_tensor': bin_tensor_2d
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family int, shape (2,) — shape (2,) 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
dotdict Ran
ran on a synthesized input · constructed an object; no result was computed
srsohn/msgi/grid-world/graph/ILP.py
code served (permissive licence) · get_code("11384de9305bcd32")
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 an instance of its own class, dotdict: the run constructed the object and did not call it, so no result was computed. 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)

class dotdict(dict):
    """dot.notation access to dictionary attributes"""
    __getattr__ = dict.get
    __setattr__ = dict.__setitem__
    __delattr__ = dict.__delitem__

# The dotdict class inherits from dict, so it takes the same arguments as dict
# We can construct it with various valid dict initialization patterns

ARGS = {
    # dotdict can be initialized like a regular dict with keyword arguments
    # or with another dict, or with key-value pairs
}

# Create an instance to verify it works
test_instance = dotdict(name="test", value=42, data=[1, 2, 3])
test_instance.new_attr = "accessible via dot notation"
retrieved = test_instance.name

print(f"dotdict created successfully: {test_instance}")
print(f"Dot notation access works: {test_instance.name}")
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 dotdict — the run constructed an instance of its own class and did not call it; no shape 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

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 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
dotdict
brick of this paper's assembly: helper
2 other papers
2 later · 1 only as an assembly brick
2006.05467 (2020-06)
Pruning neural networks without any data by iteratively conserving synaptic flow
2402.11729 (2024-02)
Prospector Heads: Generalized Feature Attribution for Large Models & Data

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

We propose and address a novel few-shot RL problem, where a task is characterized by a subtask graph which describes a set of subtasks and their dependencies that are unknown to the agent. The agent needs to quickly adapt to the task over few episodes during adaptation phase to maximize the return in the test phase. Instead of directly learning a meta-policy, we develop a Meta-learner with Subtask Graph Inference (MSGI), which infers the latent parameter of the task by interacting with the environment and maximizes the return given the latent parameter. To facilitate learning, we adopt an intrinsic reward inspired by upper confidence bound (UCB) that encourages efficient exploration. Our experiment results on two grid-world domains and StarCraft II environments show that the proposed method is able to accurately infer the latent task parameter, and to adapt more efficiently than existing meta RL and hierarchical RL methods 1 .

For agents

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

get_harvested_code_for_paper("2001.00248")
get_code_for_paper("2001.00248")
have("2001.00248")

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

curl https://syntology.ai/api/ran/2001.00248.json

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

[![Syntology run record](https://syntology.ai/api/ran/2001.00248.svg)](https://syntology.ai/paper/2001.00248)

Connect an agent — have() is free.