GPTQ model commit
Browse files- inference.py +65 -0
inference.py
ADDED
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
|
3 |
+
import argparse
|
4 |
+
|
5 |
+
def generate_prompt(question, prompt_file="prompt.md", metadata_file="metadata.sql"):
|
6 |
+
with open(prompt_file, "r") as f:
|
7 |
+
prompt = f.read()
|
8 |
+
|
9 |
+
with open(metadata_file, "r") as f:
|
10 |
+
table_metadata_string = f.read()
|
11 |
+
|
12 |
+
prompt = prompt.format(
|
13 |
+
user_question=question, table_metadata_string=table_metadata_string
|
14 |
+
)
|
15 |
+
return prompt
|
16 |
+
|
17 |
+
|
18 |
+
def get_tokenizer_model(model_name):
|
19 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
20 |
+
model = AutoModelForCausalLM.from_pretrained(
|
21 |
+
model_name,
|
22 |
+
trust_remote_code=True,
|
23 |
+
torch_dtype=torch.float16,
|
24 |
+
device_map="auto",
|
25 |
+
use_cache=True,
|
26 |
+
)
|
27 |
+
return tokenizer, model
|
28 |
+
|
29 |
+
def run_inference(question, prompt_file="prompt.md", metadata_file="metadata.sql"):
|
30 |
+
tokenizer, model = get_tokenizer_model("defog/sqlcoder")
|
31 |
+
prompt = generate_prompt(question, prompt_file, metadata_file)
|
32 |
+
|
33 |
+
# make sure the model stops generating at triple ticks
|
34 |
+
eos_token_id = tokenizer.convert_tokens_to_ids(["```"])[0]
|
35 |
+
pipe = pipeline(
|
36 |
+
"text-generation",
|
37 |
+
model=model,
|
38 |
+
tokenizer=tokenizer,
|
39 |
+
max_new_tokens=300,
|
40 |
+
do_sample=False,
|
41 |
+
num_beams=5, # do beam search with 5 beams for high quality results
|
42 |
+
)
|
43 |
+
generated_query = (
|
44 |
+
pipe(
|
45 |
+
prompt,
|
46 |
+
num_return_sequences=1,
|
47 |
+
eos_token_id=eos_token_id,
|
48 |
+
pad_token_id=eos_token_id,
|
49 |
+
)[0]["generated_text"]
|
50 |
+
.split("```sql")[-1]
|
51 |
+
.split("```")[0]
|
52 |
+
.split(";")[0]
|
53 |
+
.strip()
|
54 |
+
+ ";"
|
55 |
+
)
|
56 |
+
return generated_query
|
57 |
+
|
58 |
+
if __name__ == "__main__":
|
59 |
+
# Parse arguments
|
60 |
+
parser = argparse.ArgumentParser(description="Run inference on a question")
|
61 |
+
parser.add_argument("-q","--question", type=str, help="Question to run inference on")
|
62 |
+
args = parser.parse_args()
|
63 |
+
question = args.question
|
64 |
+
print("Loading a model and generating a SQL query for answering your question...")
|
65 |
+
print(run_inference(question))
|