File size: 10,846 Bytes
62977bb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
conditions:
  - name: bm25-default
    display: BM25 original passage (k1=0.9, b=0.4)
    display-html: BM25 original passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (1a)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage --topics $topics --output $output --bm25
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.0719
            R@1K: 0.5733
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.0802
            R@1K: 0.5839
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.1357
            nDCG@10: 0.4458
            MRR@100: 0.5060
            R@100: 0.3261
            R@1K: 0.6149
  - name: bm25-augmented-default
    display: BM25 augmented passage (k1=0.9, b=0.4)
    display-html: BM25 augmented passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (1b)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-augmented --topics $topics --output $output --bm25
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.0872
            R@1K: 0.6925
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.0917
            R@1K: 0.6933
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.0977
            nDCG@10: 0.3977
            MRR@100: 0.5303
            R@100: 0.2709
            R@1K: 0.5835
  - name: bm25-rm3-default
    display: BM25+RM3 original passage (k1=0.9, b=0.4)
    display-html: BM25+RM3 original passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (1c)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage --topics $topics --output $output --bm25 --rm3
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.0630
            R@1K: 0.5947
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.0659
            R@1K: 0.6062
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.1666
            nDCG@10: 0.4455
            MRR@100: 0.5202
            R@100: 0.3499
            R@1K: 0.6616
  - name: bm25-rm3-augmented-default
    display: BM25+RM3 augmented passage (k1=0.9, b=0.4)
    display-html: BM25+RM3 augmented passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (1d)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-augmented --topics $topics --output $output --bm25 --rm3
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.0667
            R@1K: 0.6857
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.0700
            R@1K: 0.6826
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.1050
            nDCG@10: 0.3869
            MRR@100: 0.4915
            R@100: 0.2807
            R@1K: 0.6298
  - name: bm25-d2q-t5-default
    display: BM25 w/ doc2query-T5 original passage (k1=0.9, b=0.4)
    display-html: BM25 w/ doc2query-T5 original passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (2a)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-d2q-t5 --topics $topics --output $output --bm25
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.1072
            R@1K: 0.7083
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.1123
            R@1K: 0.7151
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.1874
            nDCG@10: 0.4816
            MRR@100: 0.6848
            R@100: 0.4076
            R@1K: 0.7078
  - name: bm25-d2q-t5-augmented-default
    display: BM25 w/ doc2query-T5 augmented passage (k1=0.9, b=0.4)
    display-html: BM25 w/ doc2query-T5 augmented passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (2b)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-augmented-d2q-t5 --topics $topics --output $output --bm25
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.1172
            R@1K: 0.7647
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.1170
            R@1K: 0.7659
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.1649
            nDCG@10: 0.4702
            MRR@100: 0.6391
            R@100: 0.3883
            R@1K: 0.6962
  - name: bm25-rm3-d2q-t5-default
    display: BM25+RM3 w/ doc2query-T5 original passage (k1=0.9, b=0.4)
    display-html: BM25+RM3 w/ doc2query-T5 original passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (2c)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-d2q-t5-docvectors --topics $topics --output $output --bm25 --rm3
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.0947
            R@1K: 0.7181
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.0984
            R@1K: 0.7222
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.2285
            nDCG@10: 0.5098
            MRR@100: 0.6548
            R@100: 0.4499
            R@1K: 0.7537
  - name: bm25-rm3-d2q-t5-augmented-default
    display: BM25+RM3 w/ doc2query-T5 augmented passage (k1=0.9, b=0.4)
    display-html: BM25+RM3 w/ doc2query-T5 augmented passage (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: (2d)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-augmented-d2q-t5-docvectors --topics $topics --output $output --bm25 --rm3
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.0883
            R@1K: 0.7607
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.0904
            R@1K: 0.7649
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.1930
            nDCG@10: 0.4812
            MRR@100: 0.5958
            R@100: 0.4321
            R@1K: 0.7672
  - name: unicoil
    display: "uniCOIL (w/ doc2query-T5): pre-encoded"
    display-html: "uniCOIL (w/ doc2query-T5): pre-encoded queries"
    display-row: (3b)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-unicoil-0shot --topics $topics --output $output --hits 1000 --impact
    topics:
      - topic_key: msmarco-v2-passage-dev-unicoil
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.1499
            R@1K: 0.7616
      - topic_key: msmarco-v2-passage-dev2-unicoil
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.1577
            R@1K: 0.7671
      - topic_key: dl21-unicoil
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.2538
            nDCG@10: 0.6159
            MRR@100: 0.7311
            R@100: 0.4731
            R@1K: 0.7551
  - name: unicoil-otf
    display: "uniCOIL (w/ doc2query-T5): query inference with PyTorch"
    display-html: "uniCOIL (w/ doc2query-T5): query inference with PyTorch"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-unicoil-0shot --topics $topics --encoder castorini/unicoil-msmarco-passage --output $output --hits 1000 --impact
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.1501
            R@1K: 0.7613
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.1576
            R@1K: 0.7676
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.2539
            nDCG@10: 0.6160
            MRR@100: 0.7311
            R@100: 0.4723
            R@1K: 0.7560
  - name: unicoil-noexp
    display: "uniCOIL (noexp): pre-encoded"
    display-html: "uniCOIL (noexp): pre-encoded queries"
    display-row: (3a)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-unicoil-noexp-0shot --topics $topics --output $output --hits 1000 --impact
    topics:
      - topic_key: msmarco-v2-passage-dev-unicoil-noexp
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.1342
            R@1K: 0.7010
      - topic_key: msmarco-v2-passage-dev2-unicoil-noexp
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.1385
            R@1K: 0.7114
      - topic_key: dl21-unicoil-noexp
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.2193
            nDCG@10: 0.5756
            MRR@100: 0.6991
            R@100: 0.4246
            R@1K: 0.6897
  - name: unicoil-noexp-otf
    display: "uniCOIL (noexp): query inference with PyTorch"
    display-html: "uniCOIL (noexp): query inference with PyTorch"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v2-passage-unicoil-noexp-0shot --topics $topics --encoder castorini/unicoil-noexp-msmarco-passage --output $output --hits 1000 --impact
    topics:
      - topic_key: msmarco-v2-passage-dev
        eval_key: msmarco-v2-passage-dev
        scores:
          - MRR@100: 0.1343
            R@1K: 0.7010
      - topic_key: msmarco-v2-passage-dev2
        eval_key: msmarco-v2-passage-dev2
        scores:
          - MRR@100: 0.1385
            R@1K: 0.7114
      - topic_key: dl21
        eval_key: dl21-passage
        scores:
          - MAP@100: 0.2194
            nDCG@10: 0.5759
            MRR@100: 0.6991
            R@100: 0.4247
            R@1K: 0.6893