0.24 Load in 8bit

0.24 Log to nvidia-smi

0.24 Set device map to auto. remove eval

0.23 Set device map to cuda

0.22 remove subprocess flash attn. Set device map balanced

0.21 remove subprocess flash attn. Force device map

0.20 subprocess implementing flash_attn

0.19 implementing flash_attn

0.18 bfloat to float

0.17 changed requirements

0.14 explictly set GPU and dtype

0.14 set dtype of input_ids

0.13 catch more exceptions

0.12 catch exceptions

0.11 simplifying wo pharia

0.10 removing flash attn.

0.9 logging template func

0.8 fixed wrong device

0.7 fixing case typo pharia to Pharia

0.6 defining chat template for pharia

0.5 tweaking output, ltr

0.4 adding new example prompt ltr

0.3 resolve dependcy fckp

0.2 adding cloned space

initial commit

Duplicate from gradio-templates/chatbot