import torch from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline

1. Detect device (GPU or CPU)

DEVICE = 0 if torch.cuda.is_available() else -1

2. Load tokenizer & model from Hugging Face Hub

MODEL_NAME = "kshitijthakkar/loggenix_general"

tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, trust_remote_code=True) model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, trust_remote_code=True)

3. Set up a text-generation pipeline (optional but convenient)

generator = pipeline( "text-generation", model=model, tokenizer=tokenizer, device=DEVICE, # GPU if available, else CPU framework="pt" )

4. Simple generate function

def generate_text(prompt: str, max_length: int = 100, num_return_sequences: int = 1): output = generator( prompt, max_length=max_length, num_return_sequences=num_return_sequences, do_sample=True, # enables sampling top_k=50, # top-k sampling top_p=0.95, # nucleus sampling temperature=0.7 # adjust creativity ) return [item['generated_text'] for item in output]

5. Example usage

if name == "main": prompt = input("Enter your prompt: ") outputs = generate_text(prompt, max_length=512) for i, text in enumerate(outputs, 3): print(f"----- Generated Sequence {i} -----") print(text) print()