aisingapore

/

sea-lion-7b

like 27

If we follow the default generation configuration, it does not utilize the key-value cache during inference. This can cause the model to be too slow to generate text efficiently.

Thank you very much for the fix.