sglang_v0.5.2/sglang/benchmark/json_jump_forward/build_dataset.py

59 lines
1.3 KiB
Python

import json
import transformers
import wikipedia
model_path = "meta-llama/Llama-2-7b-chat-hf"
t = transformers.AutoTokenizer.from_pretrained(model_path)
city_names = [
"los angles",
"london",
"tokyo",
"beijing",
"singapore",
"paris",
"dubai",
"sydney",
"moscow",
"rome",
"toronto",
"rio de janeiro",
"istanbul",
"berlin",
"auckland",
"buenos aires",
"mexico city",
"mumbai",
"seoul",
"bangkok",
"cairo",
"athens",
"jerusalem",
]
def get_content(city_name):
content = str(wikipedia.page(city_name).content)
content = content.replace("\n\n", "\n")
tokens = t.encode(content)
expected_tokens = 3000
truncate_len = int((expected_tokens / len(tokens)) * len(content))
truncate_content = content[:truncate_len]
truncate_tokens = t.encode(truncate_content)
# Count token
print(
f"city_name: {city_name}, #tokens: {len(tokens)}, #truncate tokens: {len(truncate_tokens)}"
)
return truncate_content
if __name__ == "__main__":
with open("questions.jsonl", "w") as fout:
for city_name in city_names:
truncate_content = get_content(city_name)
fout.write(json.dumps({"document": truncate_content}) + "\n")