train_setfit_model_twitter_financial_news_sentiment_synthetic.py

from huggingface_hub import hf_hub_download
from datasets import load_dataset
from anyclassifier.llm.llm_client import LlamaCppClient
from anyclassifier.schema import Label
from anyclassifier import train_anyclassifier
from setfit import SetFitModel


HF_HANDLE = "user_id"


dataset = load_dataset("zeroshot/twitter-financial-news-sentiment")

llm_client = LlamaCppClient(hf_hub_download(
    "lmstudio-community/Meta-Llama-3.1-8B-Instruct-GGUF", "Meta-Llama-3.1-8B-Instruct-Q8_0.gguf"))
# or llm_client = OpenAIClient()

trainer = train_anyclassifier(
    "Classify sentiment of finance-related tweets.",
    [
        Label(id=0, desc='Bearish'),
        Label(id=1, desc='Bullish'),
        Label(id=2, desc='Neutral')
    ],
    llm_client,
    column_mapping={"text": "text"},
    model_type="setfit",
    n_record_to_generate=60,
    num_epochs=5,
    push_dataset_to_hub=True,
    is_dataset_private=True,
    metric="f1",
    metric_kwargs={"average": "micro"},
    dataset_repo_id=f"{HF_HANDLE}/test_twitter_financial_news_syn"
)
full_test_data = dataset["validation"]

print(trainer.evaluate(full_test_data))

trainer.push_to_hub(f"{HF_HANDLE}/setfit_test_twitter_news_syn", private=True)

model = SetFitModel.from_pretrained(f"{HF_HANDLE}/setfit_test_twitter_news_syn")
# Run inference
text = ["$GM - GM loses a bull https://t.co/tdUfG5HbXy",
        "Canada Goose upgraded to outperform from neutral at Baird, price target C$53"]
preds = model.predict(text)
print(text)
print(preds)