SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.11197 · NeurIPS · 2021

Teachable Reinforcement Learning via Advice Distillation

U Berkeley, Abhishek Gupta, Pieter Abbeel, Jacob Andreas, Trevor Darrell, Olivia Watkins

arXiv · PDF · Open in the Atlas

Code that ran

2 of 5 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; 2 failures of our instrument; 3 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/2203.11197.json · badge: /api/ran/2203.11197.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
rll-research/teachable canonical (classifier label)
classifier basis: classifier_2025
2 ran · 3 no recorded run
FunctionStatusWhere it lives
args_type Ran
ran; our contract draft was wrong, not the code
rll-research/teachable/scripts/train_model.py
code served (permissive licence) · get_code("02df4c3b30c85421")
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 function. 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 pathlib
import torch

torch.manual_seed(0)

# Analyze the function: args_type takes a 'default' parameter
# The function checks the type of default and returns appropriate type converters
# We need to test all branches: bool, int, pathlib.Path, None, and generic type

ARGS = {
    'default': True  # This will trigger the bool branch
}

# Additional test cases could be:
# 'default': 42  # int branch
# 'default': pathlib.Path('/tmp')  # pathlib.Path branch
# 'default': None  # None branch
# 'default': 3.14  # generic type (float) branch

# Using bool as the primary test case
ARGS = {
    'default': True
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
merge_dictlists Ran
ran; our contract draft was wrong, not the code
rll-research/teachable/algos/hierarchical_ppo_torch.py
code served (permissive licence) · get_code("9dda1e4a30cb7bf0")
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 DictList. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import torch

torch.manual_seed(0)

class DictList(dict):
    """A dictionary of lists of same size. Dictionary items can be
    accessed using `.` notation and list items using `[]` notation.

    Example:
        >>> d = DictList({"a": [[1, 2], [3, 4]], "b": [[5], [6]]})
        >>> d.a
        [[1, 2], [3, 4]]
        >>> d[0]
        DictList({"a": [1, 2], "b": [5]})
    """
    __setattr__ = dict.__setitem__

    def __getattr__(self, index):
        try:
            return dict.__getitem__(self, index)
        except KeyError:
            raise AttributeError(index)

    def __len__(self):
        return len(next(iter(dict.values(self))))

    def __getitem__(self, index):
        return DictList({key: value[index] for key, value in dict.items(self)})

    def __setitem__(self, index, d):
        try:
            for key, value in d.items():
                dict.__getitem__(self, key)[index] = value
        except:
            setattr(self, index, d)


# Create test DictList objects with different value types
dictlist1 = DictList({
    "list_field": [[1, 2], [3, 4]],
    "tensor_field": torch.randn(2, 4),
    "ndarray_field": np.array([[1.0, 2.0], [3.0, 4.0]]),
    "nested_dictlist": DictList({
        "inner_list": [[5], [6]],
        "inner_tensor": torch.randn(2, 3)
    })
})

dictlist2 = DictList({
    "list_field": [[5, 6], [7, 8]],
    "tensor_field": torch.randn(2, 4),
    "ndarray_field": np.array([[5.0, 6.0], [7.0, 8.0]]),
    "nested_dictlist": DictList({
        "inner_list": [[9], [10]],
        "inner_tensor": torch.randn(2, 3)
    })
})

list_of_dictlists = [dictlist1, dictlist2]

ARGS = {
    "list_of_dictlists": list_of_dictlists
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DictList — a DictList output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DictList 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
Agent Not yet run
unverified (harvested, no recorded run)
rll-research/teachable/algos/hierarchical_ppo_torch.py
code served (permissive licence) · get_code("18dcf30be508ff6e")
How it ran

no invocation recorded.

HierarchicalPPOAgent Not yet run
unverified (harvested, no recorded run)
rll-research/teachable/algos/hierarchical_ppo_torch.py
code served (permissive licence) · get_code("2977092da45c2a0b")
How it ran

no invocation recorded.

PPOAgent Not yet run
unverified (harvested, no recorded run)
rll-research/teachable/algos/hierarchical_ppo_torch.py
code served (permissive licence) · get_code("ac8c4379eeb797a1")
How it ran

no invocation recorded.

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 (1 with at least one paper dated earlier, 0 the same month, 0 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
merge_dictlists
brick of this paper's assembly: helper
1 other paper
1 dated earlier
1611.02779 (2016-11)
RL$^2$: Fast Reinforcement Learning via Slow Reinforcement Learning

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

Training automated agents to complete complex tasks in interactive environments is challenging: reinforcement learning requires careful hand-engineering of reward functions, imitation learning requires specialized infrastructure and access to a human expert, and learning from intermediate forms of supervision (like binary preferences) is time-consuming and extracts little information from each human intervention. Can we overcome these challenges by building agents that learn from rich, interactive feedback instead? We propose a new supervision paradigm for interactive learning based on "teachable" decision-making systems that learn from structured advice provided by an external teacher. We begin by formalizing a class of human-in-the-loop decision making problems in which multiple forms of teacher-provided advice are available to a learner. We then describe a simple learning algorithm for these problems that first learns to interpret advice, then learns from advice to complete tasks even in the absence of human supervision. In puzzle-solving, navigation, and locomotion domains, we show that agents that learn from advice can acquire new skills with significantly less human supervision than standard reinforcement learning algorithms and often less than imitation learning. 35th Conference on Neural Information Processing Systems (NeurIPS 2021), virtual.

For agents

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

get_harvested_code_for_paper("2203.11197")
get_code_for_paper("2203.11197")
have("2203.11197")

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

curl https://syntology.ai/api/ran/2203.11197.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.11197.svg)](https://syntology.ai/paper/2203.11197)

Connect an agent — have() is free.