Update README.md
Browse files
README.md
CHANGED
@@ -85,7 +85,7 @@ print(results)
|
|
85 |
# Output:
|
86 |
# SearchResults(
|
87 |
# ids=tensor([[0, 1, 2]], device='cuda:0'),
|
88 |
-
# scores=tensor([[
|
89 |
# )
|
90 |
|
91 |
query_id = 0
|
@@ -98,6 +98,7 @@ print(top1_psg)
|
|
98 |
|
99 |
```
|
100 |
|
|
|
101 |
## Building Bag-of-token Index for Search
|
102 |
|
103 |
Our framework supports using tokenization as an index (i.e., a bag-of-token index), which operates on CPU and reduces indexing time and storage requirements by over 90%, compare to an embedding-based index.
|
@@ -122,7 +123,7 @@ print(results)
|
|
122 |
# Output:
|
123 |
# SearchResults(
|
124 |
# ids=tensor([0, 2, 1], device='cuda:3'),
|
125 |
-
# scores=tensor([
|
126 |
# )
|
127 |
```
|
128 |
|
|
|
85 |
# Output:
|
86 |
# SearchResults(
|
87 |
# ids=tensor([[0, 1, 2]], device='cuda:0'),
|
88 |
+
# scores=tensor([[61.5432, 10.3108, 8.6709]], device='cuda:0')
|
89 |
# )
|
90 |
|
91 |
query_id = 0
|
|
|
98 |
|
99 |
```
|
100 |
|
101 |
+
|
102 |
## Building Bag-of-token Index for Search
|
103 |
|
104 |
Our framework supports using tokenization as an index (i.e., a bag-of-token index), which operates on CPU and reduces indexing time and storage requirements by over 90%, compare to an embedding-based index.
|
|
|
123 |
# Output:
|
124 |
# SearchResults(
|
125 |
# ids=tensor([0, 2, 1], device='cuda:3'),
|
126 |
+
# scores=tensor([61.5432, 10.3108, 8.6709], device='cuda:0')
|
127 |
# )
|
128 |
```
|
129 |
|