Views
No views yet
transformers library:1import torch
2from transformers import AutoModelForCausalLM, AutoTokenizer
3
4
5MODEL_NAME = "UUFO-Aigis/Pico-OpenLAiNN-500M" #Replace 100M with 250M or 500M if you prefer those models.
6
7tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
8model = AutoModelForCausalLM.from_pretrained(MODEL_NAME)
9
10def generate_text(prompt, model, tokenizer, max_length=512, temperature=1, top_k=50, top_p=0.95):
11 inputs = tokenizer.encode(prompt, return_tensors="pt")
12
13 outputs = model.generate(
14 inputs,
15 max_length=max_length,
16 temperature=temperature,
17 top_k=top_k,
18 top_p=top_p,
19 do_sample=True
20 )
21
22
23 generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
24 return generated_text
25
26def main():
27 # Define your prompt
28 prompt = "According to all known laws of aviation, there is no way a bee should be able to fly."
29
30 generated_text = generate_text(prompt, model, tokenizer)
31
32 print(generated_text)
33
34if __name__ == "__main__":
35 main()| Tasks | Value | Stderr | |
|---|---|---|---|
| arc_challenge | 0.1903 | ± | 0.115 |
| arc_easy | 0.4617 | ± | 0.0102 |
| boolq | 0.6034 | ± | 0.0086 |
| hellaswag | 0.3400 | ± | 0.0047 |
| lambada_openai | 0.3670 | ± | 0.0067 |
| piqa | 0.6795 | ± | 0.0109 |
| winogrande | 0.4925 | ± | 0.0141 |