update readme
Browse files
README.md
CHANGED
|
@@ -1,3 +1,23 @@
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
| 3 |
---
|
| 4 |
+
|
| 5 |
+
# FiD model trained on NQ
|
| 6 |
+
|
| 7 |
+
-- This is the model checkpoint of FiD, based on T5 large (with 770M parameters) and trained on the NQ dataset.
|
| 8 |
+
|
| 9 |
+
-- Hyperparameters: 8 x 40GB A100 GPUs; batch size 8; AdamW; LR 3e-5; 50000 steps
|
| 10 |
+
|
| 11 |
+
References:
|
| 12 |
+
[1] Natural Questions: A Benchmark for Question Answering Research. TACL 2019.
|
| 13 |
+
[2] Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering. EACL 2021.
|
| 14 |
+
|
| 15 |
+
## Model performance
|
| 16 |
+
|
| 17 |
+
We evaluate it on the NQ dataset, the EM score is 51.3 (0.1 lower than original performance reported in the paper).
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
<a href="https://huggingface.co/exbert/?model=bert-base-uncased">
|
| 21 |
+
<img width="300px" src="https://cdn-media.huggingface.co/exbert/button.png">
|
| 22 |
+
</a>
|
| 23 |
+
|