Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from datasets import load_dataset
|
2 |
+
|
3 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
4 |
+
|
5 |
+
dataset = load_dataset("CarperAI/openai_summarize_tldr")
|
6 |
+
|
7 |
+
val_prompts = [sample["prompt"] for sample in dataset["valid"]]
|
8 |
+
|
9 |
+
kwargs = {
|
10 |
+
"max_new_tokens": 50,
|
11 |
+
"do_sample": True,
|
12 |
+
"top_k": 0,
|
13 |
+
"top_p": 0.95,
|
14 |
+
"temperature": 0.5
|
15 |
+
}
|
16 |
+
|
17 |
+
model = AutoModelForCausalLM.from_pretrained("pvduy/ppo_pythia6B_sample")
|
18 |
+
model.eval()
|
19 |
+
tokenizer = AutoTokenizer.from_pretrained("pvduy/ppo_pythia6B_sample")
|
20 |
+
tokenizer.pad_token_id = tokenizer.eos_token_id
|
21 |
+
|
22 |
+
count = 0
|
23 |
+
|
24 |
+
for prompt in val_prompts:
|
25 |
+
output_tk = tokenizer(prompt, return_tensors="pt")
|
26 |
+
outputs = model.generate(output_tk.input_ids, attention_mask=output_tk.attention_mask, **kwargs)
|
27 |
+
print("Prompt:", prompt)
|
28 |
+
print("Output:", tokenizer.decode(outputs[0], skip_special_tokens=True).split("TL;DR:")[1].strip())
|
29 |
+
print("=================================")
|
30 |
+
count += 1
|
31 |
+
if count == 10:
|
32 |
+
break
|