starsnatched/
thinkergemma:latest

26 1 year ago

Trained on my Thinker dataset to replicate the thought traces of OpenAI's o1. Very smol model, very nice.

... /
params
2490e7468436 · 65B
{
"stop": [
"<start_of_turn>",
"<end_of_turn>"
]
}