Update README.md

README.md

@@ -1,566 +1,42 @@
----
-license: apache-2.0
-dataset_info:
-  features:
-  - name: text
-    dtype: string
-  - name: id
-    dtype: string
-  - name: dump
-    dtype: string
-  - name: url
-    dtype: string
-  - name: date
-    dtype: string
-  - name: file_path
-    dtype: string
-  - name: language
-    dtype: string
-  - name: language_score
-    dtype: float64
-  - name: token_count
-    dtype: int64
-  - name: line_quality
-    sequence: float64
-  splits:
-  - name: shard_0
-    num_bytes: 35429357
-    num_examples: 10000
-  - name: shard_1
-    num_bytes: 36333614
-    num_examples: 10000
-  - name: shard_2
-    num_bytes: 36121544
-    num_examples: 10000
-  - name: shard_3
-    num_bytes: 35555886
-    num_examples: 10000
-  - name: shard_4
-    num_bytes: 35217736
-    num_examples: 10000
-  - name: shard_5
-    num_bytes: 35857804
-    num_examples: 10000
-  - name: shard_6
-    num_bytes: 34950430
-    num_examples: 10000
-  - name: shard_7
-    num_bytes: 35313725
-    num_examples: 10000
-  - name: shard_8
-    num_bytes: 35159571
-    num_examples: 10000
-  - name: shard_9
-    num_bytes: 35687592
-    num_examples: 10000
-  - name: shard_10
-    num_bytes: 36790003
-    num_examples: 10000
-  - name: shard_11
-    num_bytes: 35743280
-    num_examples: 10000
-  - name: shard_12
-    num_bytes: 34898333
-    num_examples: 10000
-  - name: shard_13
-    num_bytes: 36052048
-    num_examples: 10000
-  - name: shard_14
-    num_bytes: 35283370
-    num_examples: 10000
-  - name: shard_15
-    num_bytes: 35672660
-    num_examples: 10000
-  - name: shard_16
-    num_bytes: 34919120
-    num_examples: 10000
-  - name: shard_17
-    num_bytes: 35134237
-    num_examples: 10000
-  - name: shard_18
-    num_bytes: 35038152
-    num_examples: 10000
-  - name: shard_19
-    num_bytes: 36691808
-    num_examples: 10000
-  - name: shard_20
-    num_bytes: 36025926
-    num_examples: 10000
-  - name: shard_21
-    num_bytes: 35537708
-    num_examples: 10000
-  - name: shard_22
-    num_bytes: 35097650
-    num_examples: 10000
-  - name: shard_23
-    num_bytes: 34399492
-    num_examples: 10000
-  - name: shard_24
-    num_bytes: 35352725
-    num_examples: 10000
-  - name: shard_25
-    num_bytes: 36034530
-    num_examples: 10000
-  - name: shard_26
-    num_bytes: 34819067
-    num_examples: 10000
-  - name: shard_27
-    num_bytes: 35194811
-    num_examples: 10000
-  - name: shard_28
-    num_bytes: 35788895
-    num_examples: 10000
-  - name: shard_29
-    num_bytes: 34594918
-    num_examples: 10000
-  - name: shard_30
-    num_bytes: 35010166
-    num_examples: 10000
-  - name: shard_31
-    num_bytes: 34627272
-    num_examples: 10000
-  - name: shard_32
-    num_bytes: 36228458
-    num_examples: 10000
-  - name: shard_33
-    num_bytes: 34989064
-    num_examples: 10000
-  - name: shard_34
-    num_bytes: 36037955
-    num_examples: 10000
-  - name: shard_35
-    num_bytes: 35847348
-    num_examples: 10000
-  - name: shard_36
-    num_bytes: 35367044
-    num_examples: 10000
-  - name: shard_37
-    num_bytes: 34409380
-    num_examples: 10000
-  - name: shard_38
-    num_bytes: 35376927
-    num_examples: 10000
-  - name: shard_39
-    num_bytes: 36335791
-    num_examples: 10000
-  - name: shard_40
-    num_bytes: 36112626
-    num_examples: 10000
-  - name: shard_41
-    num_bytes: 36136305
-    num_examples: 10000
-  - name: shard_42
-    num_bytes: 34834749
-    num_examples: 10000
-  - name: shard_43
-    num_bytes: 35366623
-    num_examples: 10000
-  - name: shard_44
-    num_bytes: 35501895
-    num_examples: 10000
-  - name: shard_45
-    num_bytes: 34027832
-    num_examples: 10000
-  - name: shard_46
-    num_bytes: 34234292
-    num_examples: 10000
-  - name: shard_47
-    num_bytes: 34817398
-    num_examples: 10000
-  - name: shard_48
-    num_bytes: 35798019
-    num_examples: 10000
-  - name: shard_49
-    num_bytes: 34523732
-    num_examples: 10000
-  - name: shard_50
-    num_bytes: 34755793
-    num_examples: 10000
-  - name: shard_51
-    num_bytes: 35078741
-    num_examples: 10000
-  - name: shard_52
-    num_bytes: 35634970
-    num_examples: 10000
-  - name: shard_53
-    num_bytes: 35966664
-    num_examples: 10000
-  - name: shard_54
-    num_bytes: 35287883
-    num_examples: 10000
-  - name: shard_55
-    num_bytes: 35619791
-    num_examples: 10000
-  - name: shard_56
-    num_bytes: 34561953
-    num_examples: 10000
-  - name: shard_57
-    num_bytes: 34747274
-    num_examples: 10000
-  - name: shard_58
-    num_bytes: 35428195
-    num_examples: 10000
-  - name: shard_59
-    num_bytes: 35909434
-    num_examples: 10000
-  - name: shard_60
-    num_bytes: 35745423
-    num_examples: 10000
-  - name: shard_61
-    num_bytes: 35810107
-    num_examples: 10000
-  - name: shard_62
-    num_bytes: 34290321
-    num_examples: 10000
-  - name: shard_63
-    num_bytes: 35224931
-    num_examples: 10000
-  - name: shard_64
-    num_bytes: 35321915
-    num_examples: 10000
-  - name: shard_65
-    num_bytes: 35783374
-    num_examples: 10000
-  - name: shard_66
-    num_bytes: 35955984
-    num_examples: 10000
-  - name: shard_67
-    num_bytes: 34500685
-    num_examples: 10000
-  - name: shard_68
-    num_bytes: 35574325
-    num_examples: 10000
-  - name: shard_69
-    num_bytes: 35523855
-    num_examples: 10000
-  - name: shard_70
-    num_bytes: 35518316
-    num_examples: 10000
-  - name: shard_71
-    num_bytes: 35745772
-    num_examples: 10000
-  - name: shard_72
-    num_bytes: 35571402
-    num_examples: 10000
-  - name: shard_73
-    num_bytes: 34737094
-    num_examples: 10000
-  - name: shard_74
-    num_bytes: 36056630
-    num_examples: 10000
-  - name: shard_75
-    num_bytes: 35432170
-    num_examples: 10000
-  - name: shard_76
-    num_bytes: 34261439
-    num_examples: 10000
-  - name: shard_77
-    num_bytes: 36007012
-    num_examples: 10000
-  - name: shard_78
-    num_bytes: 34440428
-    num_examples: 10000
-  - name: shard_79
-    num_bytes: 35523327
-    num_examples: 10000
-  - name: shard_80
-    num_bytes: 35363271
-    num_examples: 10000
-  - name: shard_81
-    num_bytes: 35990851
-    num_examples: 10000
-  - name: shard_82
-    num_bytes: 34833931
-    num_examples: 10000
-  - name: shard_83
-    num_bytes: 35634103
-    num_examples: 10000
-  - name: shard_84
-    num_bytes: 35550883
-    num_examples: 10000
-  - name: shard_85
-    num_bytes: 35261705
-    num_examples: 10000
-  - name: shard_86
-    num_bytes: 36258603
-    num_examples: 10000
-  - name: shard_87
-    num_bytes: 35135570
-    num_examples: 10000
-  - name: shard_88
-    num_bytes: 34759076
-    num_examples: 10000
-  - name: shard_89
-    num_bytes: 34692206
-    num_examples: 10000
-  - name: shard_90
-    num_bytes: 34247571
-    num_examples: 10000
-  - name: shard_91
-    num_bytes: 35961204
-    num_examples: 10000
-  - name: shard_92
-    num_bytes: 35020813
-    num_examples: 10000
-  - name: shard_93
-    num_bytes: 35974188
-    num_examples: 10000
-  - name: shard_94
-    num_bytes: 35047934
-    num_examples: 10000
-  - name: shard_95
-    num_bytes: 34910272
-    num_examples: 10000
-  - name: shard_96
-    num_bytes: 34996178
-    num_examples: 10000
-  - name: shard_97
-    num_bytes: 35882072
-    num_examples: 10000
-  - name: shard_98
-    num_bytes: 34902504
-    num_examples: 10000
-  - name: shard_99
-    num_bytes: 34354987
-    num_examples: 10000
-  - name: shard_100
-    num_bytes: 35295860
-    num_examples: 10000
-  - name: shard_101
-    num_bytes: 35268760
-    num_examples: 10000
-  - name: shard_102
-    num_bytes: 35249951
-    num_examples: 10000
-  - name: shard_103
-    num_bytes: 35540511
-    num_examples: 10000
-  - name: shard_104
-    num_bytes: 36118312
-    num_examples: 10000
-  - name: shard_105
-    num_bytes: 35800005
-    num_examples: 10000
-  - name: shard_106
-    num_bytes: 35325203
-    num_examples: 10000
-  download_size: 2569266812
-  dataset_size: 3783670600
-configs:
-- config_name: default
-  data_files:
-  - split: shard_0
-    path: data/shard_0-*
-  - split: shard_1
-    path: data/shard_1-*
-  - split: shard_2
-    path: data/shard_2-*
-  - split: shard_3
-    path: data/shard_3-*
-  - split: shard_4
-    path: data/shard_4-*
-  - split: shard_5
-    path: data/shard_5-*
-  - split: shard_6
-    path: data/shard_6-*
-  - split: shard_7
-    path: data/shard_7-*
-  - split: shard_8
-    path: data/shard_8-*
-  - split: shard_9
-    path: data/shard_9-*
-  - split: shard_10
-    path: data/shard_10-*
-  - split: shard_11
-    path: data/shard_11-*
-  - split: shard_12
-    path: data/shard_12-*
-  - split: shard_13
-    path: data/shard_13-*
-  - split: shard_14
-    path: data/shard_14-*
-  - split: shard_15
-    path: data/shard_15-*
-  - split: shard_16
-    path: data/shard_16-*
-  - split: shard_17
-    path: data/shard_17-*
-  - split: shard_18
-    path: data/shard_18-*
-  - split: shard_19
-    path: data/shard_19-*
-  - split: shard_20
-    path: data/shard_20-*
-  - split: shard_21
-    path: data/shard_21-*
-  - split: shard_22
-    path: data/shard_22-*
-  - split: shard_23
-    path: data/shard_23-*
-  - split: shard_24
-    path: data/shard_24-*
-  - split: shard_25
-    path: data/shard_25-*
-  - split: shard_26
-    path: data/shard_26-*
-  - split: shard_27
-    path: data/shard_27-*
-  - split: shard_28
-    path: data/shard_28-*
-  - split: shard_29
-    path: data/shard_29-*
-  - split: shard_30
-    path: data/shard_30-*
-  - split: shard_31
-    path: data/shard_31-*
-  - split: shard_32
-    path: data/shard_32-*
-  - split: shard_33
-    path: data/shard_33-*
-  - split: shard_34
-    path: data/shard_34-*
-  - split: shard_35
-    path: data/shard_35-*
-  - split: shard_36
-    path: data/shard_36-*
-  - split: shard_37
-    path: data/shard_37-*
-  - split: shard_38
-    path: data/shard_38-*
-  - split: shard_39
-    path: data/shard_39-*
-  - split: shard_40
-    path: data/shard_40-*
-  - split: shard_41
-    path: data/shard_41-*
-  - split: shard_42
-    path: data/shard_42-*
-  - split: shard_43
-    path: data/shard_43-*
-  - split: shard_44
-    path: data/shard_44-*
-  - split: shard_45
-    path: data/shard_45-*
-  - split: shard_46
-    path: data/shard_46-*
-  - split: shard_47
-    path: data/shard_47-*
-  - split: shard_48
-    path: data/shard_48-*
-  - split: shard_49
-    path: data/shard_49-*
-  - split: shard_50
-    path: data/shard_50-*
-  - split: shard_51
-    path: data/shard_51-*
-  - split: shard_52
-    path: data/shard_52-*
-  - split: shard_53
-    path: data/shard_53-*
-  - split: shard_54
-    path: data/shard_54-*
-  - split: shard_55
-    path: data/shard_55-*
-  - split: shard_56
-    path: data/shard_56-*
-  - split: shard_57
-    path: data/shard_57-*
-  - split: shard_58
-    path: data/shard_58-*
-  - split: shard_59
-    path: data/shard_59-*
-  - split: shard_60
-    path: data/shard_60-*
-  - split: shard_61
-    path: data/shard_61-*
-  - split: shard_62
-    path: data/shard_62-*
-  - split: shard_63
-    path: data/shard_63-*
-  - split: shard_64
-    path: data/shard_64-*
-  - split: shard_65
-    path: data/shard_65-*
-  - split: shard_66
-    path: data/shard_66-*
-  - split: shard_67
-    path: data/shard_67-*
-  - split: shard_68
-    path: data/shard_68-*
-  - split: shard_69
-    path: data/shard_69-*
-  - split: shard_70
-    path: data/shard_70-*
-  - split: shard_71
-    path: data/shard_71-*
-  - split: shard_72
-    path: data/shard_72-*
-  - split: shard_73
-    path: data/shard_73-*
-  - split: shard_74
-    path: data/shard_74-*
-  - split: shard_75
-    path: data/shard_75-*
-  - split: shard_76
-    path: data/shard_76-*
-  - split: shard_77
-    path: data/shard_77-*
-  - split: shard_78
-    path: data/shard_78-*
-  - split: shard_79
-    path: data/shard_79-*
-  - split: shard_80
-    path: data/shard_80-*
-  - split: shard_81
-    path: data/shard_81-*
-  - split: shard_82
-    path: data/shard_82-*
-  - split: shard_83
-    path: data/shard_83-*
-  - split: shard_84
-    path: data/shard_84-*
-  - split: shard_85
-    path: data/shard_85-*
-  - split: shard_86
-    path: data/shard_86-*
-  - split: shard_87
-    path: data/shard_87-*
-  - split: shard_88
-    path: data/shard_88-*
-  - split: shard_89
-    path: data/shard_89-*
-  - split: shard_90
-    path: data/shard_90-*
-  - split: shard_91
-    path: data/shard_91-*
-  - split: shard_92
-    path: data/shard_92-*
-  - split: shard_93
-    path: data/shard_93-*
-  - split: shard_94
-    path: data/shard_94-*
-  - split: shard_95
-    path: data/shard_95-*
-  - split: shard_96
-    path: data/shard_96-*
-  - split: shard_97
-    path: data/shard_97-*
-  - split: shard_98
-    path: data/shard_98-*
-  - split: shard_99
-    path: data/shard_99-*
-  - split: shard_100
-    path: data/shard_100-*
-  - split: shard_101
-    path: data/shard_101-*
-  - split: shard_102
-    path: data/shard_102-*
-  - split: shard_103
-    path: data/shard_103-*
-  - split: shard_104
-    path: data/shard_104-*
-  - split: shard_105
-    path: data/shard_105-*
-  - split: shard_106
-    path: data/shard_106-*
----
+---
+license: apache-2.0
+---
+
+# Dataset Card for FinerWeb-10BT
+
+## Dataset Details
+
+### Dataset Description
+
+This dataset extends the FineWeb-10BT sample (10 billion tokens) by adding quality scores for each line of text. Each document has been enhanced with line-level quality scores derived from an LLM-based filtering pipeline that identifies high and low-quality content.
+
+- **Curated by:** Erik Henriksson*, Otto Tarkka*, Filip Ginter (University of Turku, *Equal contribution.)
+- **Language(s):** English
+- **License:** apache-2.0
+
+### Dataset Sources
+- **Repository:** https://huggingface.co/datasets/TurkuNLP/finerweb-10bt
+- **Paper:** Coming soon!
+
+## Dataset Structure
+
+The dataset follows the original FineWeb-10BT structure with an additional `line_quality` key for each document. This key contains a list of floating-point scores (0.0 to 1.0) corresponding to each line in the document (obtained by splitting the document's text on newlines). Higher scores indicate higher quality content, with scores closer to 1.0 representing clean, natural language text, and lower scores indicating content like formatting artifacts, copyright notices, or navigation elements.
+
+## Dataset Creation
+
+### Source Data
+
+#### Data Collection and Processing
+
+Quality scores were generated through a pipeline that:
+1. Used GPT-4o mini to label a 20,000-document sample
+2. Trained a DeBERTa-v3 classifier on the labeled data
+3. Applied the classifier to generate quality scores for each line in the full dataset
+
+## Bias, Risks, and Limitations
+
+The quality scores inherit some biases from the LLMs used in the labeling process. Users should note that the distinction between high and low-quality content can be subjective, and the scores should be interpreted as guidelines rather than absolute measures.
+
+## Citation
+
+Coming soon!