Hugging Face's logo

Datasets:
zaaabik
/
alien-data

License:
Dataset card Data Studio Files
xet
 Community
Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code:   DatasetGenerationError
Exception:    ArrowInvalid
Message:      Schema at index 50 was different: 
model: string
adapter: string
dataset: string
train_on_dataset: string
normalization: bool
seed: int64
test-acc: double
md_roc_auc: double
md_marginal_roc_auc: double
md_relative_roc_auc: double
test-base-max-prob-roc-auc: double
rde_n_components_256_roc_auc: double
vs
model: string
adapter: string
dataset: string
train_on_dataset: string
normalization: bool
seed: int64
test-acc: double
md_roc_auc: double
md_marginal_roc_auc: double
md_relative_roc_auc: double
test-base-max-prob-roc-auc: double
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1848, in _prepare_split_single
                  writer.write_table(table)
                  ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 783, in write_table
                  self.write_rows_on_file()  # in case there are buffered rows to write first
                  ~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 662, in write_rows_on_file
                  table = pa.concat_tables(self.current_rows)
                File "pyarrow/table.pxi", line 6268, in pyarrow.lib.concat_tables
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                  return check_status(status)
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: Schema at index 50 was different: 
              model: string
              adapter: string
              dataset: string
              train_on_dataset: string
              normalization: bool
              seed: int64
              test-acc: double
              md_roc_auc: double
              md_marginal_roc_auc: double
              md_relative_roc_auc: double
              test-base-max-prob-roc-auc: double
              rde_n_components_256_roc_auc: double
              vs
              model: string
              adapter: string
              dataset: string
              train_on_dataset: string
              normalization: bool
              seed: int64
              test-acc: double
              md_roc_auc: double
              md_marginal_roc_auc: double
              md_relative_roc_auc: double
              test-base-max-prob-roc-auc: double
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1868, in _prepare_split_single
                  num_examples, num_bytes = writer.finalize()
                                            ~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 803, in finalize
                  self.write_rows_on_file()
                  ~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 662, in write_rows_on_file
                  table = pa.concat_tables(self.current_rows)
                File "pyarrow/table.pxi", line 6268, in pyarrow.lib.concat_tables
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                  return check_status(status)
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: Schema at index 50 was different: 
              model: string
              adapter: string
              dataset: string
              train_on_dataset: string
              normalization: bool
              seed: int64
              test-acc: double
              md_roc_auc: double
              md_marginal_roc_auc: double
              md_relative_roc_auc: double
              test-base-max-prob-roc-auc: double
              rde_n_components_256_roc_auc: double
              vs
              model: string
              adapter: string
              dataset: string
              train_on_dataset: string
              normalization: bool
              seed: int64
              test-acc: double
              md_roc_auc: double
              md_marginal_roc_auc: double
              md_relative_roc_auc: double
              test-base-max-prob-roc-auc: double
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1879, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

model
string
adapter
string
dataset
string
train_on_dataset
string
normalization
bool
seed
int64
test-acc
float64
md_roc_auc
float64
md_marginal_roc_auc
float64
md_relative_roc_auc
float64
test-base-max-prob-roc-auc
float64
rde_n_components_256_roc_auc
float64
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/CoLa/0/
full_trained
cola
train
true
0
0.854267
0.812275
0.812371
0.654705
0.785431
0.74777
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/CoLa/1/
full_trained
cola
train
true
1
0.837967
0.829404
0.829459
0.668551
0.786285
0.760991
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/CoLa/2/
full_trained
cola
train
true
2
0.837967
0.809818
0.809886
0.650407
0.806325
0.763882
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/CoLa/3/
full_trained
cola
train
true
3
0.846596
0.808855
0.80894
0.615381
0.791064
0.661467
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/CoLa/4/
full_trained
cola
train
true
4
0.840844
0.799858
0.7999
0.629906
0.817388
0.802029
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/NewsGroups/0/
full_trained
20newsgroups
train
true
0
0.781333
0.805379
0.803189
0.821503
0.838859
0.589965
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/NewsGroups/1/
full_trained
20newsgroups
train
true
1
0.774695
0.813823
0.811212
0.829925
0.851305
0.80696
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/NewsGroups/2/
full_trained
20newsgroups
train
true
2
0.77496
0.81659
0.813705
0.842163
0.863255
0.720671
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/NewsGroups/3/
full_trained
20newsgroups
train
true
3
0.775358
0.812193
0.808951
0.842973
0.857043
0.72761
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/NewsGroups/4/
full_trained
20newsgroups
train
true
4
0.779209
0.803323
0.800726
0.822507
0.839157
0.744106
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST2/0/
full_trained
SST2
train
true
0
0.946101
0.878014
0.877911
0.73372
0.856273
0.536789
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST2/1/
full_trained
SST2
train
true
1
0.927752
0.885691
0.885593
0.825103
0.91105
0.89505
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST2/2/
full_trained
SST2
train
true
2
0.941514
0.884932
0.884837
0.813714
0.880323
0.895966
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST2/3/
full_trained
SST2
train
true
3
0.938073
0.863715
0.863556
0.798266
0.886908
0.524835
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST2/4/
full_trained
SST2
train
true
4
0.938073
0.885855
0.885787
0.822195
0.887168
0.853957
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST5/0/
full_trained
SST5
train
true
0
0.537557
0.55461
0.554259
0.574816
0.596026
0.570954
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST5/1/
full_trained
SST5
train
true
1
0.541176
0.568111
0.566467
0.567012
0.570262
0.558323
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST5/2/
full_trained
SST5
train
true
2
0.547059
0.542228
0.541154
0.586123
0.576962
0.541337
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST5/3/
full_trained
SST5
train
true
3
0.556109
0.562587
0.562622
0.541318
0.554978
0.558272
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/SST5/4/
full_trained
SST5
train
true
4
0.555204
0.530057
0.529753
0.55943
0.611673
0.550142
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/ToxigenDataset/0/
full_trained
toxigen
train
true
0
0.82234
0.732871
0.732855
0.610306
0.707795
0.726201
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/ToxigenDataset/1/
full_trained
toxigen
train
true
1
0.811702
0.729073
0.729043
0.646985
0.745933
0.727873
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/ToxigenDataset/2/
full_trained
toxigen
train
true
2
0.807447
0.730941
0.730956
0.638831
0.7539
0.741642
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/ToxigenDataset/3/
full_trained
toxigen
train
true
3
0.820213
0.709015
0.709031
0.616114
0.711072
0.713022
/workspace/AdUE/data/full-fine-tune-models/electra_runs_seeds/ToxigenDataset/4/
full_trained
toxigen
train
true
4
0.814894
0.763745
0.763693
0.672271
0.73355
0.548198
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/CoLa/0/
full_trained
cola
train
true
0
0.837009
0.77425
0.77419
0.644293
0.758635
0.690098
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/CoLa/1/
full_trained
cola
train
true
1
0.819751
0.694616
0.693981
0.656212
0.793701
0.597838
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/CoLa/2/
full_trained
cola
train
true
2
0.834132
0.775852
0.775932
0.656618
0.775972
0.722354
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/CoLa/3/
full_trained
cola
train
true
3
0.814957
0.76769
0.767708
0.630165
0.787662
0.704928
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/CoLa/4/
full_trained
cola
train
true
4
0.831256
0.784396
0.784487
0.675248
0.774097
0.775299
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/NewsGroups/0/
full_trained
20newsgroups
train
true
0
0.780536
0.835093
0.833712
0.845068
0.840224
0.744635
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/NewsGroups/1/
full_trained
20newsgroups
train
true
1
0.767127
0.835709
0.834072
0.853671
0.844058
0.786357
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/NewsGroups/2/
full_trained
20newsgroups
train
true
2
0.783192
0.83055
0.829496
0.829593
0.835282
0.779231
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/NewsGroups/3/
full_trained
20newsgroups
train
true
3
0.774695
0.847934
0.846287
0.853315
0.851741
0.752248
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/NewsGroups/4/
full_trained
20newsgroups
train
true
4
0.774562
0.847713
0.845561
0.867223
0.85784
0.823965
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST2/0/
full_trained
SST2
train
true
0
0.93578
0.84445
0.844297
0.72735
0.804469
0.843466
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST2/1/
full_trained
SST2
train
true
1
0.949541
0.818703
0.818566
0.697354
0.78383
0.50936
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST2/2/
full_trained
SST2
train
true
2
0.940367
0.852392
0.852017
0.742706
0.853881
0.852134
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST2/3/
full_trained
SST2
train
true
3
0.940367
0.847444
0.84728
0.776032
0.811796
0.838977
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST2/4/
full_trained
SST2
train
true
4
0.938073
0.852101
0.852124
0.708503
0.816479
0.802047
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST5/0/
full_trained
SST5
train
true
0
0.557014
0.560071
0.559771
0.578378
0.606056
0.571088
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST5/1/
full_trained
SST5
train
true
1
0.562443
0.557247
0.556899
0.574313
0.590879
0.562146
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST5/2/
full_trained
SST5
train
true
2
0.547511
0.551003
0.550707
0.566143
0.626176
0.552359
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST5/3/
full_trained
SST5
train
true
3
0.553846
0.587303
0.586744
0.587372
0.587137
0.586259
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/SST5/4/
full_trained
SST5
train
true
4
0.540724
0.604227
0.603938
0.571014
0.590152
0.603091
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/ToxigenDataset/0/
full_trained
toxigen
train
true
0
0.804255
0.758519
0.758555
0.674848
0.736675
0.594419
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/ToxigenDataset/1/
full_trained
toxigen
train
true
1
0.8
0.617976
0.617516
0.627893
0.700264
0.586726
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/ToxigenDataset/2/
full_trained
toxigen
train
true
2
0.794681
0.589266
0.588912
0.639338
0.704625
0.562485
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/ToxigenDataset/3/
full_trained
toxigen
train
true
3
0.817021
0.717153
0.717009
0.643865
0.754183
0.733792
/workspace/AdUE/data/full-fine-tune-models/roberta_runs_seeds/ToxigenDataset/4/
full_trained
toxigen
train
true
4
0.823404
0.747541
0.747533
0.648088
0.727503
0.745634
meta-llama/Llama-2-7b-chat-hf
base_full
arcc_5_shot
train
true
0
0.575085
0.5
0.482497
0.517503
0.613037
null
meta-llama/Llama-2-7b-chat-hf
base_full
HellaSwag_0_shot
train
true
0
0.472316
0.5
0.502852
0.497148
0.703695
null
meta-llama/Llama-2-7b-chat-hf
base_full
HellaSwag_5_shot
train
true
0
0.362179
0.5
0.37007
0.62993
0.713366
null
meta-llama/Llama-2-7b-chat-hf
base_full
MMLU_0_shot
train
true
0
0.463538
0.5
0.504114
0.495886
0.69524
null
meta-llama/Llama-2-7b-chat-hf
base_full
MMLU_5_shot
train
true
0
0.42964
0.5
0.501035
0.498965
0.689728
null
meta-llama/Llama-2-7b-chat-hf
base_full
truthfulqa_1_shot
train
true
0
0.287805
0.5
0.472254
0.527746
0.56919
null
meta-llama/Llama-2-7b-chat-hf
base_full
truthfulqa_5_shot
train
true
0
0.302439
0.5
0.459283
0.540717
0.624295
null
Qwen/Qwen2.5-7B-Instruct
base_full
arcc_5_shot
train
true
0
0.890785
0.5
0.476667
0.523333
0.773168
null
Qwen/Qwen2.5-7B-Instruct
base_full
HellaSwag_0_shot
train
true
0
0.881597
0.5
0.579844
0.420156
0.741621
null
Qwen/Qwen2.5-7B-Instruct
base_full
HellaSwag_5_shot
train
true
5
0.883788
0.5
0.597536
0.402464
0.80798
null
Qwen/Qwen2.5-7B-Instruct
base_full
MMLU_0_shot
train
true
0
0.673551
0.5
0.503965
0.496035
0.818577
null
Qwen/Qwen2.5-7B-Instruct
base_full
MMLU_5_shot
train
true
0
0.729383
0.5
0.494164
0.505836
0.753436
null
Qwen/Qwen2.5-7B-Instruct
base_full
truthfulqa_1_shot
train
true
0
0.64878
0.5
0.548768
0.451232
0.701546
null
Qwen/Qwen2.5-7B-Instruct
base_full
truthfulqa_5_shot
train
true
0
0.64878
0.5
0.459064
0.540936
0.809106
null
null
null
truthfulqa_5_shot
train
null
0
null
null
null
null
null
null
null
null
truthfulqa_5_shot
train
null
0
null
null
null
null
null
null
null
null
truthfulqa_5_shot
train
null
0
null
null
null
null
null
null
null
null
truthfulqa_5_shot
train
null
0
null
null
null
null
null
null
null
null
truthfulqa_5_shot
train
null
0
null
null
null
null
null
null
null
null
truthfulqa_5_shot
train
null
0
null
null
null
null
null
null
null
null
cola
train
null
0
null
null
null
null
null
null
null
null
cola
train
null
0
null
null
null
null
null
null
null
null
cola
train
null
0
null
null
null
null
null
null
null
null
cola
train
null
1
null
null
null
null
null
null
null
null
cola
train
null
1
null
null
null
null
null
null
null
null
cola
train
null
1
null
null
null
null
null
null
null
null
cola
train
null
2
null
null
null
null
null
null
null
null
cola
train
null
2
null
null
null
null
null
null
null
null
cola
train
null
2
null
null
null
null
null
null
null
null
cola
train
null
3
null
null
null
null
null
null
null
null
cola
train
null
3
null
null
null
null
null
null
null
null
cola
train
null
3
null
null
null
null
null
null
null
null
cola
train
null
4
null
null
null
null
null
null
null
null
cola
train
null
4
null
null
null
null
null
null
null
null
cola
train
null
4
null
null
null
null
null
null
null
null
SST2
train
null
0
null
null
null
null
null
null
null
null
SST2
train
null
0
null
null
null
null
null
null
null
null
SST2
train
null
0
null
null
null
null
null
null
null
null
SST2
train
null
1
null
null
null
null
null
null
null
null
SST2
train
null
1
null
null
null
null
null
null
null
null
SST2
train
null
1
null
null
null
null
null
null
null
null
SST2
train
null
2
null
null
null
null
null
null
null
null
SST2
train
null
2
null
null
null
null
null
null
null
null
SST2
train
null
2
null
null
null
null
null
null
null
null
SST2
train
null
3
null
null
null
null
null
null
null
null
SST2
train
null
3
null
null
null
null
null
null
null
null
SST2
train
null
3
null
null
null
null
null
null
null
null
SST2
train
null
4
null
null
null
null
null
null
null
null
SST2
train
null
4
null
null
null
null
null
null
null
null
SST2
train
null
4
null
null
null
null
null
null
End of preview.

ALIEN — data and results

Data, model adapters and experiment outputs for ALIEN: Aligned Entropy Head for Improving Uncertainty Estimation of LLMs (submitted results + revision experiments). Code: the ALIEN repository, revision/ folder.

Layout

folder content
results/ Published runs used for the paper tables: results_on_embeddings/grid_* (ALIEN, its variants and ablations; best_run.pickle = test scores + metric_df with the selected hyper-parameters), baseline_results_on_embeddings_rde/ (MD, RMD, MDM, RDE), probes_ds_with_lr_hyper/ (linear / attention-pooling probes), full-fine-tuning and prompted-LLM blocks
data/embeddings/ Pooled train/val/test features, logits, targets and the classification head per task model: <model>_results_ds_<seed>_train_gird_full/LoraConfig_<dataset>_state.pickle
data/models/ LoRA adapters of the task models: <model>_runs_seeds/<Dataset>/<seed>/
revision_runs/heldout/head_seeds/ Base model seed 0: ALIEN, ConfidNet, last-layer linear probe re-trained with up to 20 head seeds (hyper-parameters fixed per model × dataset)
revision_runs/heldout/head_seeds_base{1..4}/ Same for base model seeds 1–4 (5 head seeds); *_own_hp/ = runs with each base seed's own hyper-parameters
revision_runs/heldout/alien, confidnet, embeddings_heldout Held-out-error protocol pilot (heads trained on 85 % of the validation split)
revision_runs/probes_seeds/ Linear + attention-pooling probes (last / mid / begin layer) with head seeds: model_<m>/dataset_<D>/base_<k>/<probe>_layer_<L>/head_seed_<s>.pt, lr.json = learning rate per probe
revision_runs/outputs/ Aggregated tables (per-seed metrics, paired tests, calibration, selective risk)

Reproduce the paper metrics

With the ALIEN repository checked out and this dataset downloaded to DATA:

python revision/scripts/reproduce_paper_tables.py --results-dir DATA/results --out-dir out --seed-dirs seed_0
python revision/scripts/compare_with_paper.py     --results-dir DATA/results --out-dir out
python revision/scripts/make_selection_tables.py  --results-dir DATA/results --out-dir out --tables-dir out/tables

Seed-based statistics: revision/scripts/summarize_nested.py --work DATA/revision_runs/heldout --results-dir DATA/results.
Downloads last month
163
Total file size:
22.7 GB