model_inference.py

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

MODEL_NAME = "NousResearch/Hermes-2-Pro-Mistral"  # ✅ Uncensored & efficient

tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, use_fast=True)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_NAME,
    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
    device_map="auto"
)
model.eval()

def generate_code(prompt: str, max_tokens: int = 256):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    with torch.no_grad():
        output = model.generate(
            inputs.input_ids,
            max_new_tokens=max_tokens,
            do_sample=True,
            temperature=0.7,
            top_p=0.95,
            repetition_penalty=1.1,
            pad_token_id=tokenizer.eos_token_id
        )
    return tokenizer.decode(output[0], skip_special_tokens=True)