SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2210.09337 · NeurIPS · 2022

Robust Imitation of a Few Demonstrations with a Backwards Model

Jung Park, Lawson Wong

arXiv · PDF · Open in the Atlas

Code that ran

4 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 3 have no contract verdict; 1 failure of our instrument; 1 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/2210.09337.json · badge: /api/ran/2210.09337.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
jypark0/bmil canonical (classifier label)
classifier basis not recorded on the link
4 ran · 1 no recorded run
FunctionStatusWhere it lives
get_logger Ran
ran; our contract draft was wrong, not the code
jypark0/bmil/src/logger.py
code served (permissive licence) · get_code("edca5c1880cc6ad4")
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 Logger. 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 logging

ARGS = {
    "name": "test_logger",
    "level": logging.DEBUG
}
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 Logger — a Logger output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Logger 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
load_demos Ran
ran on a synthesized input
jypark0/bmil/src/demo_utils.py
code served (permissive licence) · get_code("43f8666cc61c5a4e")
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 dict. 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 pickle
import numpy as np
import tempfile
import os

# Create a temporary pickle file with demo data for the Push environment
def create_demo_file(env_type):
    # Create demo data structure
    demo = {
        'obs': [],
        'obs_next': [],
        'actions': [],
        'rewards': [],
        'dones': []
    }
    
    # Create 2 episodes
    for e in range(2):
        demo['obs'].append([])
        demo['obs_next'].append([])
        demo['actions'].append([])
        demo['rewards'].append([])
        demo['dones'].append([])
        
        # Create 3 timesteps per episode
        for t in range(3):
            # obs and obs_next are dicts with 'observation', 'achieved_goal', 'desired_goal'
            obs_dict = {
                'observation': np.random.randn(4),
                'achieved_goal': np.random.randn(3),
                'desired_goal': np.random.randn(3)
            }
            obs_next_dict = {
                'observation': np.random.randn(4),
                'achieved_goal': np.random.randn(3),
                'desired_goal': np.random.randn(3)
            }
            
            demo['obs'][e].append(obs_dict)
            demo['obs_next'][e].append(obs_next_dict)
            demo['actions'][e].append(np.random.randn(2))
            demo['rewards'][e].append(0.0)
            demo['dones'][e].append(False)
    
    # Write to temporary file
    temp_fd, temp_path = tempfile.mkstemp(suffix='.pkl')
    os.close(temp_fd)
    
    with open(temp_path, 'wb') as f:
        pickle.dump(demo, f)
    
    return temp_path

# Create demo file for Push environment
demo_file = create_demo_file('Push')

ARGS = {
    'filename': demo_file,
    'env_id': 'PushEnv-v0'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
threshold_linear_fn Ran
ran on a synthesized input
jypark0/bmil/src/utils.py
code served (permissive licence) · get_code("ce7a08b9c4e1b28b")
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 torch

torch.manual_seed(0)

ARGS = {
    "current_epoch": 5,
    "epoch_range": [0, 10],
    "value_range": [10, 100]
}
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 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
to_torch Ran
ran on a synthesized input
jypark0/bmil/src/utils.py
code served (permissive licence) · get_code("aebc04c1ce6a0e50")
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 (2, 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)

# Create a small numpy array
array = np.random.randn(2, 4).astype(np.float32)

# Use CPU device
device = 'cpu'

ARGS = {
    'array': array,
    'device': device
}
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 (2, 4) — shape (2, 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
to_np Not yet run
unverified (harvested, no recorded run)
jypark0/bmil/src/utils.py
code served (permissive licence) · get_code("3294140a01532925")
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, 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
get_logger 3 other papers
1 dated earlier, 1 later, 1 same year, order not determinable · 1 only as an assembly brick
1312.5602 (2013-12)
Playing Atari with Deep Reinforcement Learning
2022.findings-emnlp.242 (2022 (year only))
ER-TEST: Evaluating Explanation Regularization Methods for Language Models
2407.08751 (2024-07)
Latent Diffusion for Neural Spiking 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

Behavior cloning of expert demonstrations can speed up learning optimal policies in a more sample-efficient way over reinforcement learning. However, the policy cannot extrapolate well to unseen states outside of the demonstration data, creating covariate shift (agent drifting away from demonstrations) and compounding errors. In this work, we tackle this issue by extending the region of attraction around the demonstrations so that the agent can learn how to get back onto the demonstrated trajectories if it veers off-course. We train a generative backwards dynamics model and generate short imagined trajectories from states in the demonstrations. By imitating both demonstrations and these model rollouts, the agent learns the demonstrated paths and how to get back onto these paths. With optimal or nearoptimal demonstrations, the learned policy will be both optimal and robust to deviations, with a wider region of attraction. On continuous control domains, we evaluate the robustness when starting from different initial states unseen in the demonstration data. While both our method and other imitation learning baselines can successfully solve the tasks for initial states in the training distribution, our method exhibits considerably more robustness to different initial states.

For agents

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

get_harvested_code_for_paper("2210.09337")
get_code_for_paper("2210.09337")
have("2210.09337")

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

curl https://syntology.ai/api/ran/2210.09337.json

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

[![Syntology run record](https://syntology.ai/api/ran/2210.09337.svg)](https://syntology.ai/paper/2210.09337)

Connect an agent — have() is free.