- Training hardware: 4 AMD MI200
- Batch size: 128
- Samples per query: 512 (8 * 4 * 16)
- Data: msmarco-passage 4K
- Learning steps: 4K with 0.4K warm-up
- Learning rate: 1e-4
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
๐
Ask for provider support