train_setfit.py

import json
from datasets import load_dataset
from setfit import SetFitModel, SetFitTrainer
from huggingface_hub import upload_file

LABELS = ["pre-1900","1900–1945","1946–1990","1991–2008","2009–2015","2016–2018","2019–2022","2023–present"]
name2id = {n:i for i,n in enumerate(LABELS)}

ds = load_dataset("json", data_files={"train":"train.jsonl","val":"val.jsonl"})
# Assert all labels are present in train
seen = set([row["label"] for row in ds["train"]])
assert seen.issuperset(LABELS), f"Train set missing labels: {set(LABELS)-seen}"

ds = ds.map(lambda x: {"label": name2id[x["label"]]})

model = SetFitModel.from_pretrained("sentence-transformers/paraphrase-mpnet-base-v2", num_labels=len(LABELS))
trainer = SetFitTrainer(
    model=model,
    train_dataset=ds["train"],
    eval_dataset=ds["val"],
    metric="accuracy",
    num_iterations=20,
    num_epochs=2,
    batch_size=16
)
trainer.train()
print("Eval:", trainer.evaluate())

# push model
repo_id = "DelaliScratchwerk/text-period-setfit"
trainer.push_to_hub(repo_id)
print("Pushed to:", repo_id)

# push labels.json alongside the model
with open("labels.json","w") as f:
    json.dump(LABELS, f)
upload_file(path_or_fileobj="labels.json", path_in_repo="labels.json", repo_id=repo_id, repo_type="model")
print("Uploaded labels.json")