12 4 months ago

A very small test: gemma3:1b fine tuned with a dataset created with gemma3:27b from an article by Laurent Nottale on Scale Relativity. This is pretty much useless, i did it only to learn to generate a dataset and fine tune a small model.

3116c5225075 · 77B
{
"stop": [
"<end_of_turn>"
],
"temperature": 1,
"top_k": 64,
"top_p": 0.95
}