Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -19,6 +19,12 @@ configs:
|
|
19 |
path: pair-score/train-*
|
20 |
- split: test
|
21 |
path: pair-score/test-*
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
- config_name: triplet
|
23 |
data_files:
|
24 |
- split: train
|
@@ -75,4 +81,13 @@ We filter all rows where negative scores exceed any of the similarity scores.
|
|
75 |
The 4 instruction types produce results of varying similarity scores,
|
76 |
with the most similar being `synonym` and least similar `contextual meaning`.
|
77 |
|
78 |
-
<img src="cdf_plot_scores.png" alt="CDF Plot" width="600"/>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
path: pair-score/train-*
|
20 |
- split: test
|
21 |
path: pair-score/test-*
|
22 |
+
- config_name: pair-score-hard
|
23 |
+
data_files:
|
24 |
+
- split: train
|
25 |
+
path: pair-score-hard/train-*
|
26 |
+
- split: test
|
27 |
+
path: pair-score-hard/test-*
|
28 |
- config_name: triplet
|
29 |
data_files:
|
30 |
- split: train
|
|
|
81 |
The 4 instruction types produce results of varying similarity scores,
|
82 |
with the most similar being `synonym` and least similar `contextual meaning`.
|
83 |
|
84 |
+
<img src="cdf_plot_scores.png" alt="CDF Plot" width="600"/>
|
85 |
+
|
86 |
+
|
87 |
+
## Subsets
|
88 |
+
|
89 |
+
* `pair-score` - random choice weighted to a target of 0.9
|
90 |
+
* `pair-score-hard` random choice weighted to a target of 0.85
|
91 |
+
* `triplet` - random choice weighted to a target of 0.9
|
92 |
+
* `triplet-hard` - random choice weighted to a target of 0.85
|
93 |
+
* `raw` - full dataset
|