| # Example usage | |
| ```python | |
| from transformers import AutoTokenizer, AutoModelForCausalLM | |
| import transformers | |
| import torch | |
| model = "tiiuae/falcon-40B" | |
| tokenizer = AutoTokenizer.from_pretrained(model) | |
| pipeline = transformers.pipeline( | |
| "text-generation", | |
| model=model, | |
| tokenizer=tokenizer, | |
| torch_dtype=torch.bfloat16, | |
| trust_remote_code=True, | |
| device_map="auto", | |
| ) | |
| sequences = pipeline( | |
| "My favourite dad joke is", | |
| max_length=200, | |
| do_sample=True, | |
| top_k=10, | |
| num_return_sequences=1, | |
| eos_token_id=tokenizer.eos_token_id, | |
| ) | |
| print("=" * 30) | |
| print("Results:") | |
| for seq in sequences: | |
| print(f"Result: {seq['generated_text']}") | |
| ``` | |