T5-ANCE
T5-ANCE generally follows the training procedure described in this page, but uses a much larger batch size.
Dataset used for training:
- MS MARCO Passage
Evaluation result:
| Dataset | Metric | Result |
|---|---|---|
| MS MARCO Passage (dev) | MRR@10 | 0.3570 |
Important hyper-parameters:
| Name | Value |
|---|---|
| Global batch size | 256 |
| Learning rate | 5e-6 |
| Maximum length of query | 32 |
| Maximum length of document | 128 |
| Template for query | <text> |
| Template for document | Title: <title> Text: <text> |
Paper
-
- Downloads last month
- 169
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.