-
Nicolas Patry authored
* Softcapping for gemma2. * Less clutter. * No access to transformers config, only config_dict here. * 0.0 is the null value in the C++ API.
6aeb6690
* Softcapping for gemma2. * Less clutter. * No access to transformers config, only config_dict here. * 0.0 is the null value in the C++ API.