SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.05091 · ICML · 2021

PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training

Pieter Abbeel, Kimin Lee, Laura Smith

arXiv · PDF · Open in the Atlas

Code that ran

7 of 7 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 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; 4 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/2106.05091.json · badge: /api/ran/2106.05091.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/bpref canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
pokaxpoka/b_pref community
Papers with Code archive
3 ran · 0 no recorded run
FunctionStatusWhere it lives
KCenterGreedy Ran
ran; our contract draft was wrong, not the code
pokaxpoka/b_pref/reward_model.py
pointer only (licence: NONE) · get_code("c30edaf748298765")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, pokaxpoka/b_pref when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

KCenterGreedy Ran
ran; our contract draft was wrong, not the code
rll-research/bpref/reward_model.py
code served (permissive licence) · get_code("f707d589e5cb7e51")
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. 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 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 numpy as np
import torch

torch.manual_seed(0)

# Small numpy arrays for obs and full_obs
# obs: shape (6, 4) - 6 observations, 4 dimensions
# full_obs: shape (4, 4) - 4 observations, 4 dimensions
obs = np.random.randn(6, 4).astype(np.float32)
full_obs = np.random.randn(4, 4).astype(np.float32)

# Number of new samples to select
num_new_sample = 2

ARGS = {
    'obs': obs,
    'full_obs': full_obs,
    'num_new_sample': num_new_sample
}
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 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
RewardModel Ran
ran on a synthesized input · constructed an object; no result was computed
pokaxpoka/b_pref/reward_model.py
pointer only (licence: NONE) · get_code("7317e4f9684221bb")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, pokaxpoka/b_pref when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

compute_smallest_dist Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pokaxpoka/b_pref/reward_model.py
pointer only (licence: NONE) · get_code("9d7e96fcfe3ff0f4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, pokaxpoka/b_pref when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

compute_state_entropy Ran
ran on a synthesized input
rll-research/bpref/agent/sac.py
code served (permissive licence) · get_code("115ab9ed9d356045")
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, 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 small tensors for the function
# obs: shape (batch_size_obs, feature_dim)
# full_obs: shape (batch_size_full, feature_dim)
# k: integer for kth nearest neighbor

obs = torch.randn(2, 4)  # 2 observations, 4 features
full_obs = torch.randn(10, 4)  # 10 full observations, 4 features
k = 3

ARGS = {
    "obs": obs,
    "full_obs": full_obs,
    "k": k,
}
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 (2, 1) — shape (2, 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
gen_net Ran
ran; our contract draft was wrong, not the code
rll-research/bpref/reward_model.py
pointer only (licence: MIT); the stored copy get_code serves, from moonoutcloudback/clarify_pbrl, is pointer only (licence: NONE) · get_code("a17d99b8eda680ff")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from moonoutcloudback/clarify_pbrl, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

linear_schedule Ran
ran on a synthesized input
rll-research/bpref/train_PPO.py
code served (permissive licence) · get_code("ad4e19d5ac571c31")
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 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 torch
import numpy as np

torch.manual_seed(0)

# The function linear_schedule takes an initial_value parameter
# which can be either a float or a str that can be converted to float
# It returns a callable that takes progress_remaining and returns a float

# Test with a float value
initial_value_float = 0.001

ARGS = {
    "initial_value": initial_value_float
}
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 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

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 5 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 5 later) and 1 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
gen_net
brick of this paper's assembly: helper
8 other papers
8 later · 3 only as an assembly brick
2111.03026 (2021-11)
B-Pref: Benchmarking Preference-Based Reinforcement Learning
2205.12401 (2022-05)
Reward Uncertainty for Exploration in Preference-based Reinforcement Learning
2305.17400 (2023-05)
Query-Policy Misalignment in Preference-Based Reinforcement Learning
and 5 more not named here
KCenterGreedy
brick of this paper's assembly: sampler
3 other papers
3 later · 1 only as an assembly brick
2111.03026 (2021-11)
B-Pref: Benchmarking Preference-Based Reinforcement Learning
2205.12401 (2022-05)
Reward Uncertainty for Exploration in Preference-based Reinforcement Learning
2402.17257 (2024-02)
RIME: Robust Preference-based Reinforcement Learning with Noisy Preferences
compute_smallest_dist
brick of this paper's assembly: helper
3 other papers
3 later · 1 only as an assembly brick
2111.03026 (2021-11)
B-Pref: Benchmarking Preference-Based Reinforcement Learning
2205.12401 (2022-05)
Reward Uncertainty for Exploration in Preference-based Reinforcement Learning
2402.17257 (2024-02)
RIME: Robust Preference-based Reinforcement Learning with Noisy Preferences
compute_state_entropy 3 other papers
3 later
2111.03026 (2021-11)
B-Pref: Benchmarking Preference-Based Reinforcement Learning
2305.17400 (2023-05)
Query-Policy Misalignment in Preference-Based Reinforcement Learning
2506.12822 (2025-06)
Enhancing Rating-Based Reinforcement Learning to Effectively Leverage Feedback from Large Vision-Language Models
linear_schedule 1 other paper
1 later
2111.03026 (2021-11)
B-Pref: Benchmarking Preference-Based 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

Conveying complex objectives to reinforcement learning (RL) agents can often be difficult, involving meticulous design of reward functions that are sufficiently informative yet easy enough to provide. Human-in-the-loop RL methods allow practitioners to instead interactively teach agents through tailored feedback; however, such approaches have been challenging to scale since human feedback is very expensive. In this work, we aim to make this process more sample-and feedback-efficient. We present an off-policy, interactive RL algorithm that capitalizes on the strengths of both feedback and off-policy learning. Specifically, we learn a reward model by actively querying a teacher's preferences between two clips of behavior and use it to train an agent. To enable off-policy learning, we relabel all the agent's past experience when its reward model changes. We additionally show that pre-training our agents with unsupervised exploration substantially increases the mileage of its queries. We demonstrate that our approach is capable of learning tasks of higher complexity than previously considered by human-in-the-loop methods, including a variety of locomotion and robotic manipulation skills. We also show that our method is able to utilize real-time human feedback to effectively prevent reward exploitation and learn new behaviors that are difficult to specify with standard reward functions.

For agents

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

get_harvested_code_for_paper("2106.05091")
get_code_for_paper("2106.05091")
have("2106.05091")

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

curl https://syntology.ai/api/ran/2106.05091.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.05091.svg)](https://syntology.ai/paper/2106.05091)

Connect an agent — have() is free.