text2sql finetune

README.md

@@ -0,0 +1,36 @@
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+# Load the tokenizer
+tokenizer = AutoTokenizer.from_pretrained("t5-small")
+
+# Load the model
+model = AutoModelForSeq2SeqLM.from_pretrained("t5-small")
+
+# Rest of the code for preparing input, generating predictions, and decoding the output...
+
+
+from typing import List
+
+table_prefix = "table:"
+question_prefix = "question:"
+
+def prepare_input(question: str, table: List[str]):
+    print("question:", question)
+    print("table:", table)
+    join_table = ",".join(table)
+    inputs = f"{question_prefix} {question} {table_prefix} {join_table}"
+    input_ids = tokenizer(inputs, max_length=700, return_tensors="pt").input_ids
+    return input_ids
+
+def inference(question: str, table: List[str]) -> str:
+    input_data = prepare_input(question=question, table=table)
+    input_data = input_data.to(model.device)
+    outputs = model.generate(inputs=input_data, num_beams=10, top_k=10, max_length=512)
+    result = tokenizer.decode(token_ids=outputs[0], skip_special_tokens=True)
+    return result
+
+test_id = 1000
+print("model result:", inference(dataset["test"][test_id]["question"], dataset["test"][test_id]["table"]["header"]))
+print("real result:", dataset["test"][test_id]["sql"]["human_readable"])
+
+inference("what is id with name jui and age equal 25", ["id","name", "age"])