File size: 1,457 Bytes
e7ac24f f907397 e7ac24f f907397 e7ac24f f907397 e7ac24f f907397 e7ac24f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 |
# STACKS Usage Example
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
def load_stacks():
"""Load STACKS model and tokenizer"""
model = AutoModelForCausalLM.from_pretrained(
"gouthamsai78/STACKS",
torch_dtype=torch.bfloat16,
device_map="auto",
attn_implementation="eager"
)
tokenizer = AutoTokenizer.from_pretrained("gouthamsai78/STACKS")
return model, tokenizer
def generate_prompt(role, model, tokenizer, temperature=0.8):
"""Generate creative prompt for given role"""
input_text = f"### Task: Generate a creative prompt for someone acting as {role}\n### Generated Prompt:"
inputs = tokenizer(input_text, return_tensors="pt")
with torch.no_grad():
outputs = model.generate(
**inputs,
max_new_tokens=200,
temperature=temperature,
do_sample=True,
top_p=0.9,
repetition_penalty=1.1,
pad_token_id=tokenizer.eos_token_id
)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
return response[len(input_text):].strip()
# Example usage
if __name__ == "__main__":
model, tokenizer = load_stacks()
# Generate prompts for different roles
roles = ["chef", "detective", "astronaut", "teacher", "artist"]
for role in roles:
prompt = generate_prompt(role, model, tokenizer)
print(f"**{role.title()}**: {prompt}\n")
|