Datasets:

Modalities:
Tabular
ArXiv:
DOI:
License:
Dataset Viewer
Auto-converted to Parquet
Search is not available for this dataset
config_id
int64
0
255
training_steps
int64
5k
10M
performance
float64
-3,734.97
34.6k
hp_config.buffer_batch_size
int64
16
256
hp_config.buffer_prio_sampling
bool
2 classes
hp_config.buffer_size
int64
1.1k
1,000k
hp_config.initial_epsilon
float64
0.5
1
hp_config.learning_rate
float64
0
0.1
hp_config.learning_starts
int64
1
32.7k
hp_config.target_epsilon
float64
0
0.2
hp_config.use_target_network
bool
2 classes
hp_config.buffer_alpha
float64
0.02
0.98
hp_config.buffer_beta
float64
0.01
1
hp_config.buffer_epsilon
float64
0
0
hp_config.target_update_interval
float64
1
2k
seed
int64
0
9
0
1,000,000
0
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
2,000,000
1,523.4375
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
3,000,000
0
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
4,000,000
0
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
5,000,000
546.875
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
6,000,000
2,000
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
7,000,000
1,000
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
8,000,000
0
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
9,000,000
3,000
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
0
10,000,000
3,000
32
false
603,170
0.772442
0.000131
21,527
0.08808
false
null
null
null
null
9
1
1,000,000
1,351.5625
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
2,000,000
3,632.8125
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
3,000,000
2,117.1875
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
4,000,000
2,351.5625
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
5,000,000
7,984.375
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
6,000,000
4,171.875
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
7,000,000
4,828.125
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
8,000,000
7,226.5625
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
9,000,000
3,703.125
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
1
10,000,000
10,210.9375
32
false
71,987
0.543565
0.000001
27,455
0.155853
false
null
null
null
null
9
2
1,000,000
257.8125
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
2,000,000
2,937.5
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
3,000,000
4,187.5
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
4,000,000
4,453.125
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
5,000,000
1,453.125
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
6,000,000
0
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
7,000,000
2,507.8125
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
8,000,000
3,000
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
9,000,000
1,421.875
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
2
10,000,000
0
16
false
144,230
0.972334
0.000407
14,187
0.053647
false
null
null
null
null
9
3
1,000,000
4,289.0625
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
2,000,000
8,835.9375
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
3,000,000
5,578.125
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
4,000,000
2,960.9375
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
5,000,000
9,132.8125
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
6,000,000
12,875
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
7,000,000
14,265.625
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
8,000,000
16,570.312
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
9,000,000
8,593.75
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
3
10,000,000
20,132.812
32
false
943,806
0.84091
0.000063
14,897
0.139829
true
null
null
null
258
9
4
1,000,000
554.6875
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
2,000,000
2,000
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
3,000,000
609.375
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
4,000,000
3,000
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
5,000,000
2,000
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
6,000,000
0
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
7,000,000
0
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
8,000,000
0
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
9,000,000
593.75
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
4
10,000,000
531.25
16
true
570,637
0.719301
0.087472
4,263
0.042566
true
0.656577
0.260759
0.000007
489
9
5
1,000,000
2,945.3125
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
2,000,000
5,531.25
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
3,000,000
8,726.5625
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
4,000,000
5,031.25
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
5,000,000
3,914.0625
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
6,000,000
2,335.9375
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
7,000,000
8,414.0625
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
8,000,000
7,921.875
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
9,000,000
11,320.3125
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
5
10,000,000
11,765.625
16
true
656,682
0.569091
0.00001
12,729
0.164378
true
0.839565
0.105137
0.000805
938
9
6
1,000,000
2,953.125
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
2,000,000
7,429.6875
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
3,000,000
5,515.625
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
4,000,000
6,171.875
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
5,000,000
4,859.375
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
6,000,000
10,953.125
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
7,000,000
7,367.1875
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
8,000,000
17,015.625
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
9,000,000
13,718.75
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
6
10,000,000
16,390.625
64
false
739,531
0.519594
0.000026
4,839
0.059932
true
null
null
null
1,385
9
7
1,000,000
1,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
2,000,000
1,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
3,000,000
1,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
4,000,000
1,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
5,000,000
531.25
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
6,000,000
2,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
7,000,000
2,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
8,000,000
2,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
9,000,000
671.875
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
7
10,000,000
1,000
32
true
523,736
0.54697
0.000758
30,524
0.064395
false
0.14048
0.719164
0.000001
null
9
8
1,000,000
554.6875
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
2,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
3,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
4,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
5,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
6,000,000
1,812.5
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
7,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
8,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
9,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
8
10,000,000
0
32
true
829,116
0.502348
0.00245
9,595
0.147304
false
0.256266
0.580396
0.000023
null
9
9
1,000,000
890.625
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
2,000,000
4,000
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
3,000,000
3,000
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
4,000,000
0
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
5,000,000
1,460.9375
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
6,000,000
0
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
7,000,000
0
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
8,000,000
546.875
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
9,000,000
1,429.6875
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9
9
10,000,000
1,398.4375
16
false
447,691
0.923204
0.003143
10,466
0.162946
true
null
null
null
1,386
9

The ARLBench Performance Dataset

ARLBench is a benchmark designed for hyperparameter optimization (HPO) in Reinforcement Learning (RL). Given that we conducted several thousand runs to identify meaningful HPO test settings for RL, we have compiled these results into a dataset for future research and applications.

This dataset can be leveraged to:

  • Meta-learn insights about the hyperparameter landscape in RL.
  • Warm-start HPO tools by utilizing previously explored configurations.

Dataset Details

The dataset includes:

  • Landscape data: 10 runs each for PPO, DQN, and SAC across:
    • Atari-5 environments
    • Four XLand gridworlds
    • Four Brax walkers
    • Five classic control environments
    • Two Box2D environments
  • Optimization data: 3 runs per optimization algorithm for each algorithm-environment combination, covering:
    • Population-Based Training (PBT)
    • SMAC
    • SMAC with Multi-Fidelity
    • Random Search

Dataset Mapping

The dataset follows this mapping: training steps, seed, hyperparameter configurationtraining performance\text{training steps, seed, hyperparameter configuration} \mapsto \text{training performance}

For optimization runs, it additionally includes:

  • Optimization seed: Differentiates between the five optimization runs per algorithm-environment pair.
  • Optimization step: Tracks configurations evaluated at different steps.

Example Usage

You can find example notebooks demonstrating how to use:

For more details, refer to the ARLBench paper.

Downloads last month
620