- Training hardware: 2 AMD MI200
- Batch size: 64
- Samples per query: 512 (32 * 2 * 8)
- Data: msmarco-passage-titled 491K
- Learning steps: 25K with 2.5K warm-up
- Learning rate: 1e-5
- Downloads last month
- 29
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
๐
Ask for provider support