MatBar99 commited on
Commit
7493573
·
verified ·
1 Parent(s): 0f52b7f

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +153 -3
README.md CHANGED
@@ -1,14 +1,164 @@
1
  ---
2
  language:
3
  - en
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  tags:
5
  - asr
 
 
6
  - hallucination
 
7
  - whisper
8
  - earnings22
9
- pretty_name: 'HALAS: A Human-Annotated Dataset of Hallucinations of Modern ASR Systems'
10
- size_categories:
11
- - 1K<n<10K
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12
  ---
13
  # Dataset Card for HALAS
14
 
 
1
  ---
2
  language:
3
  - en
4
+
5
+ license: unknown
6
+
7
+ task_categories:
8
+ - automatic-speech-recognition
9
+ - audio-classification
10
+ - text-classification
11
+
12
+ annotations_creators:
13
+ - expert-generated
14
+
15
+ source_datasets:
16
+ - earnings22
17
+
18
+ multilinguality:
19
+ - monolingual
20
+
21
+ size_categories:
22
+ - 1K<n<10K
23
+
24
+ pretty_name: "HALAS: A Human-Annotated Dataset of Hallucinations of Modern ASR Systems"
25
+
26
  tags:
27
  - asr
28
+ - speech-recognition
29
+ - speech-processing
30
  - hallucination
31
+ - hallucination-detection
32
  - whisper
33
  - earnings22
34
+ - benchmark
35
+ - human-annotations
36
+ - error-detection
37
+ - audio
38
+ - speech-to-text
39
+ - robustness
40
+
41
+ dataset_info:
42
+ - config_name: default
43
+ features:
44
+ - name: audio_id
45
+ dtype: string
46
+
47
+ - name: audio_duration
48
+ dtype: float32
49
+
50
+ - name: e22_reference_text
51
+ dtype: string
52
+
53
+ - name: corrected_reference_text
54
+ dtype: string
55
+
56
+ - name: whisper_large_v2_prediction
57
+ dtype: string
58
+ - name: whisper_large_v2_label
59
+ dtype: bool
60
+ - name: whisper_large_v2_hallucination_text
61
+ dtype: string
62
+ - name: whisper_large_v2_hallucination_json
63
+ dtype: string
64
+
65
+ - name: whisper_large_v3_prediction
66
+ dtype: string
67
+ - name: whisper_large_v3_label
68
+ dtype: bool
69
+ - name: whisper_large_v3_hallucination_text
70
+ dtype: string
71
+ - name: whisper_large_v3_hallucination_json
72
+ dtype: string
73
+
74
+ - name: whisper_large_v3_turbo_prediction
75
+ dtype: string
76
+ - name: whisper_large_v3_turbo_label
77
+ dtype: bool
78
+ - name: whisper_large_v3_turbo_hallucination_text
79
+ dtype: string
80
+ - name: whisper_large_v3_turbo_hallucination_json
81
+ dtype: string
82
+
83
+ - name: crisper_whisper_prediction
84
+ dtype: string
85
+ - name: crisper_whisper_label
86
+ dtype: bool
87
+ - name: crisper_whisper_hallucination_text
88
+ dtype: string
89
+ - name: crisper_whisper_hallucination_json
90
+ dtype: string
91
+
92
+ - name: canary_prediction
93
+ dtype: string
94
+ - name: canary_label
95
+ dtype: bool
96
+ - name: canary_hallucination_text
97
+ dtype: string
98
+ - name: canary_hallucination_json
99
+ dtype: string
100
+
101
+ - name: canary_flash_prediction
102
+ dtype: string
103
+ - name: canary_flash_label
104
+ dtype: bool
105
+ - name: canary_flash_hallucination_text
106
+ dtype: string
107
+ - name: canary_flash_hallucination_json
108
+ dtype: string
109
+
110
+ - name: parakeet_prediction
111
+ dtype: string
112
+ - name: parakeet_label
113
+ dtype: bool
114
+ - name: parakeet_hallucination_text
115
+ dtype: string
116
+ - name: parakeet_hallucination_json
117
+ dtype: string
118
+
119
+ - name: phi4_prediction
120
+ dtype: string
121
+ - name: phi4_label
122
+ dtype: bool
123
+ - name: phi4_hallucination_text
124
+ dtype: string
125
+ - name: phi4_hallucination_json
126
+ dtype: string
127
+
128
+ - name: granite_prediction
129
+ dtype: string
130
+ - name: granite_label
131
+ dtype: bool
132
+ - name: granite_hallucination_text
133
+ dtype: string
134
+ - name: granite_hallucination_json
135
+ dtype: string
136
+
137
+ - name: split
138
+ dtype: string
139
+
140
+ splits:
141
+ - name: train
142
+ num_examples: 2866
143
+
144
+ - name: test
145
+ num_examples: 745
146
+
147
+ configs:
148
+ - config_name: default
149
+ data_files:
150
+ - split: train
151
+ path: train-*
152
+ - split: test
153
+ path: test-*
154
+
155
+ paperswithcode_id: null
156
+
157
+ train_eval_index:
158
+ - config: default
159
+ task: hallucination-detection
160
+ split: test
161
+
162
  ---
163
  # Dataset Card for HALAS
164