OLDhandler.py

from transformers import AutoModelForCausalLM, AutoTokenizer

class Handler:
    def __init__(self, model_path):
        self.model_path = model_path
        self.tokenizer = AutoTokenizer.from_pretrained(model_path)
        self.model = AutoModelForCausalLM.from_pretrained(model_path)
        self.model.eval()

    def generate_text(self, prompt, max_length=100, num_return_sequences=1, temperature=0.7):
        input_ids = self.tokenizer.encode(prompt, return_tensors="pt")

        generated_ids = self.model.generate(
            input_ids,
            max_length=max_length,
            num_return_sequences=num_return_sequences,
            temperature=temperature,
            pad_token_id=self.tokenizer.eos_token_id,
        )

        generated_texts = [self.tokenizer.decode(ids, skip_special_tokens=True) for ids in generated_ids]
        return generated_texts

    def __call__(self, request):
        # Parse the request and extract the necessary information
        prompt = request["prompt"]
        max_length = request.get("max_length", 100)
        num_return_sequences = request.get("num_return_sequences", 1)
        temperature = request.get("temperature", 0.7)

        # Generate text based on the prompt and parameters
        generated_texts = self.generate_text(prompt, max_length, num_return_sequences, temperature)

        # Prepare the response
        response = {
            "generated_texts": generated_texts
        }

        return response

handler = Handler(".")