Commit
·
24b6b76
1
Parent(s):
11979d9
updated
Browse files
README.md
CHANGED
|
@@ -4,16 +4,16 @@ tags:
|
|
| 4 |
- dpr
|
| 5 |
license: apache-2.0
|
| 6 |
model-index:
|
| 7 |
-
- name: dpr-question_encoder_bert_uncased_L-
|
| 8 |
results: []
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information Keras had access to. You should
|
| 12 |
probably proofread and complete it, then remove this comment. -->
|
| 13 |
|
| 14 |
-
# dpr-question_encoder_bert_uncased_L-
|
| 15 |
|
| 16 |
-
This model(google/bert_uncased_L-
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
|
| 19 |
|
|
@@ -23,14 +23,15 @@ evaluation dataset: facebook-dpr-dev-dataset from official DPR github
|
|
| 23 |
|
| 24 |
|model_name|data_name|num of queries|num of passages|R@10|R@20|R@50|R@100|R@100|
|
| 25 |
|---|---|---|---|---|---|---|---|---|
|
| 26 |
-
|nlpconnect/dpr-ctx_encoder_bert_uncased_L-
|
|
|
|
| 27 |
|*facebook/dpr-ctx_encoder-single-nq-base(hf/fb)|nq-dev dataset|6445|199795|40.94%|49.27%|59.05%|66.00%|82.00%|
|
| 28 |
|
| 29 |
evaluation dataset: UKPLab/beir test data but we have used first 2lac passage only.
|
| 30 |
|
| 31 |
|model_name|data_name|num of queries|num of passages|R@10|R@20|R@50|R@100|R@100|
|
| 32 |
|---|---|---|---|---|---|---|---|---|
|
| 33 |
-
|nlpconnect/dpr-ctx_encoder_bert_uncased_L-
|
| 34 |
|*facebook/dpr-ctx_encoder-single-nq-base(hf/fb)|nq-test dataset|3452|200001|32.93%|43.74%|56.95%|66.30%|83.92%|
|
| 35 |
|
| 36 |
Note: * means we have evaluated on same eval dataset.
|
|
|
|
| 4 |
- dpr
|
| 5 |
license: apache-2.0
|
| 6 |
model-index:
|
| 7 |
+
- name: dpr-question_encoder_bert_uncased_L-2_H-128_A-2
|
| 8 |
results: []
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information Keras had access to. You should
|
| 12 |
probably proofread and complete it, then remove this comment. -->
|
| 13 |
|
| 14 |
+
# dpr-question_encoder_bert_uncased_L-2_H-128_A-2
|
| 15 |
|
| 16 |
+
This model(google/bert_uncased_L-2_H-128_A-2) was trained from scratch on training data: data.retriever.nq-adv-hn-train(facebookresearch/DPR).
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
|
| 19 |
|
|
|
|
| 23 |
|
| 24 |
|model_name|data_name|num of queries|num of passages|R@10|R@20|R@50|R@100|R@100|
|
| 25 |
|---|---|---|---|---|---|---|---|---|
|
| 26 |
+
|nlpconnect/dpr-ctx_encoder_bert_uncased_L-2_H-128_A-2(our)|nq-dev dataset|6445|199795|60.53%|68.28%|76.07%|80.98%|91.45%|
|
| 27 |
+
|nlpconnect/dpr-ctx_encoder_bert_uncased_L-12_H-128_A-2(our)|nq-dev dataset|6445|199795|65.43%|71.99%|79.03%|83.24%|92.11%|
|
| 28 |
|*facebook/dpr-ctx_encoder-single-nq-base(hf/fb)|nq-dev dataset|6445|199795|40.94%|49.27%|59.05%|66.00%|82.00%|
|
| 29 |
|
| 30 |
evaluation dataset: UKPLab/beir test data but we have used first 2lac passage only.
|
| 31 |
|
| 32 |
|model_name|data_name|num of queries|num of passages|R@10|R@20|R@50|R@100|R@100|
|
| 33 |
|---|---|---|---|---|---|---|---|---|
|
| 34 |
+
|nlpconnect/dpr-ctx_encoder_bert_uncased_L-2_H-128_A-2(our)|nq-test dataset|3452|200001|49.68%|59.06%|69.40%|75.75%|89.28%|
|
| 35 |
|*facebook/dpr-ctx_encoder-single-nq-base(hf/fb)|nq-test dataset|3452|200001|32.93%|43.74%|56.95%|66.30%|83.92%|
|
| 36 |
|
| 37 |
Note: * means we have evaluated on same eval dataset.
|