I trained my own model with
openassistant-oasst1-h2ogpt dataset on kaggle bu t my model is still giving so weird answers does anyone have an idea to fix?
Siz: hello
I ’ m h2oGPT , a large language model by H2O . ai , a large language model by H2O . ai , a large language model . ai . ai , how can you help ? , how may I assist you ? . Nice ! human ? ? ! , how are you you have any questions ? , or help ? ? ! I can help ? ? , , what can do you feel ? , what ’ d me ? . you have any other ? . . I ’ m a little ? . I ’ m.
[
1.16s | 
120 token]
Siz: how are you today
i i can have what ? ’ t ’ sure , what do you mean mean to write ? ? ? ’ ’ ? . what do you tell me me ? ? ? to you tell me me ? . ? ! ! a large , my ! ! I ’ m a chat ? . < human > : I ’ m going to try ? < human . < human ? ! < human > : ! i can there the other of my help ? > . -.
[ 1.17s | 120 token]
Siz: what are you talking about
Tekrar tespit edildi, üretim durduruldu
was was a large - year . . < human > :.
[ 0.22s | 21 token]
Siz: what model are you
it ’ t have been a great idea of his age to make it . , i ’ m a love , my day day . a love and my grand , a love . i ’ m so good to meet , i ’ m just got the light . my make a great day . i ’ m a great friend . is my best to you , let ’ re sure . . I ’ m here here to you ? , we ’ ll to be you the best best of my own . < human > > : the sun of me ?
Note: h2ogpt is in dataset i think i dont know