Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -43,6 +43,26 @@ configs:
|
|
43 |
data_files:
|
44 |
- split: test
|
45 |
path: data/qrecc/test.json
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
---
|
47 |
|
48 |
## ConvRAG Bench
|
@@ -50,14 +70,31 @@ ConvRAG Bench is a benchmark for evaluating a model's conversational QA capabili
|
|
50 |
|
51 |
ConvRAG Bench also includes evaluations for the unanswerable scenario, where we evaluate models' capability to determine whether the answer to the question can be found within the given context. Equipping models with such capability can substantially decrease the likelihood of hallucination.
|
52 |
|
53 |
-
##
|
54 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
|
|
|
|
|
|
|
|
|
56 |
|
57 |
## License
|
58 |
The ConvRAG are built on and derived from existing datasets. We refer users to the original licenses accompanying each dataset.
|
59 |
|
60 |
-
|
61 |
## Citation
|
62 |
If you evaluate using ConvRAG, please cite all the datasets you use.
|
63 |
<pre>
|
|
|
43 |
data_files:
|
44 |
- split: test
|
45 |
path: data/qrecc/test.json
|
46 |
+
- config_name: doqa_cooking
|
47 |
+
data_files:
|
48 |
+
- split: test
|
49 |
+
path: data/doqa/test_cooking.json
|
50 |
+
- config_name: doqa_movies
|
51 |
+
data_files:
|
52 |
+
- split: test
|
53 |
+
path: data/doqa/test_movies.json
|
54 |
+
- config_name: doqa_travel
|
55 |
+
data_files:
|
56 |
+
- split: test
|
57 |
+
path: data/doqa/test_travel.json
|
58 |
+
- config_name: sqa
|
59 |
+
data_files:
|
60 |
+
- split: test
|
61 |
+
path: data/sqa/test.json
|
62 |
+
- config_name: convfinqa
|
63 |
+
data_files:
|
64 |
+
- split: dev
|
65 |
+
path: data/convfinqa/dev.json
|
66 |
---
|
67 |
|
68 |
## ConvRAG Bench
|
|
|
70 |
|
71 |
ConvRAG Bench also includes evaluations for the unanswerable scenario, where we evaluate models' capability to determine whether the answer to the question can be found within the given context. Equipping models with such capability can substantially decrease the likelihood of hallucination.
|
72 |
|
73 |
+
## Benchmark Results
|
74 |
+
|
75 |
+
| | ChatQA-1.0-7B | Command-R-Plus | Llama-3-instruct-70b | GPT-4-0613 | ChatQA-1.0-70B | ChatQA-1.5-8B | ChatQA-1.5-70B |
|
76 |
+
| -- |:--:|:--:|:--:|:--:|:--:|:--:|:--:|
|
77 |
+
| Doc2Dial | 37.88 | 33.51 | 37.88 | 34.16 | 38.9 | 39.33 | 41.26 |
|
78 |
+
| QuAC | 29.69 | 34.16 | 36.96 | 40.29 | 41.82 | 39.73 | 38.82 |
|
79 |
+
| QReCC | 46.97 | 49.77 | 51.34 | 52.01 | 48.05 | 49.03 | 51.40 |
|
80 |
+
| CoQA | 76.61 | 69.71 | 76.98 | 77.42 | 78.57 | 76.46 | 78.44 |
|
81 |
+
| DoQA | 41.57 | 40.67 | 41.24 | 43.39 | 51.94 | 49.6 | 50.67 |
|
82 |
+
| ConvFinQA | 51.61 | 71.21 | 76.6 | 81.28 | 73.69 | 78.46 | 81.88 |
|
83 |
+
| SQA | 61.87 | 74.07 | 69.61 | 79.21 | 69.14 | 73.28 | 83.82 |
|
84 |
+
| TopioCQA | 45.45 | 53.77 | 49.72 | 45.09 | 50.98 | 49.96 | 55.63 |
|
85 |
+
| HybriDial* | 54.51 | 46.7 | 48.59 | 49.81 | 56.44 | 65.76 | 68.27 |
|
86 |
+
| INSCIT | 30.96 | 35.76 | 36.23 | 36.34 | 31.9 | 30.1 | 32.31 |
|
87 |
+
| Average (all) | 47.71 | 50.93 | 52.52 | 53.90 | 54.14 | 55.17 | 58.25 |
|
88 |
+
| Average (exclude HybriDial) | 46.96 | 51.40 | 52.95 | 54.35 | 53.89 | 53.99 | 57.14 |
|
89 |
|
90 |
+
Note that ChatQA-1.5 used some samples from the HybriDial training dataset. To ensure fair comparison, we also compare average scores excluding HybriDial.
|
91 |
+
|
92 |
+
## Evaluation
|
93 |
+
We open-source the scripts for running and evaluating on ConvRAG (including the unanswerable evaluations).
|
94 |
|
95 |
## License
|
96 |
The ConvRAG are built on and derived from existing datasets. We refer users to the original licenses accompanying each dataset.
|
97 |
|
|
|
98 |
## Citation
|
99 |
If you evaluate using ConvRAG, please cite all the datasets you use.
|
100 |
<pre>
|