How to use OscarXZQ/logical_deduction_three_objects with PEFT:
from peft import PeftModel from transformers import AutoModelForSeq2SeqLM base_model = AutoModelForSeq2SeqLM.from_pretrained("google/flan-t5-large") model = PeftModel.from_pretrained(base_model, "OscarXZQ/logical_deduction_three_objects")
03cae36 b1dbf02 03cae36
1
2
3
4
5
6
7
8
{ "epoch": 5.0, "total_flos": 91796629155840.0, "train_loss": 0.5963952255249023, "train_runtime": 32.1448, "train_samples_per_second": 3.111, "train_steps_per_second": 3.111 }
