Create genealogy_bot.py
Browse files- genealogy_bot.py +23 -0
genealogy_bot.py
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# genealogy_bot.py
|
| 2 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
|
| 3 |
+
from peft import PeftModel
|
| 4 |
+
from config import BASE_MODEL, MODEL_DIR
|
| 5 |
+
|
| 6 |
+
class GenealogyBot:
|
| 7 |
+
def __init__(self):
|
| 8 |
+
self.tokenizer = AutoTokenizer.from_pretrained(MODEL_DIR)
|
| 9 |
+
base_model = AutoModelForCausalLM.from_pretrained(
|
| 10 |
+
BASE_MODEL,
|
| 11 |
+
torch_dtype="auto",
|
| 12 |
+
trust_remote_code=True
|
| 13 |
+
)
|
| 14 |
+
self.model = PeftModel.from_pretrained(base_model, MODEL_DIR)
|
| 15 |
+
self.model.eval()
|
| 16 |
+
self.pipe = pipeline("text-generation", model=self.model, tokenizer=self.tokenizer, max_new_tokens=100)
|
| 17 |
+
|
| 18 |
+
def query(self, question: str) -> str:
|
| 19 |
+
prompt = f"<|im_start|>user\n{question}<|im_end|>\n<|im_start|>assistant\n"
|
| 20 |
+
result = self.pipe(prompt, do_sample=False)
|
| 21 |
+
full_text = result[0]['generated_text']
|
| 22 |
+
answer = full_text.split("<|im_start|>assistant\n")[-1].split("<|im_end|>")[0].strip()
|
| 23 |
+
return answer if answer else "抱歉,无法回答该问题。"
|