Skip to content

Latest commit

 

History

History
68 lines (47 loc) · 5.55 KB

README.md

File metadata and controls

68 lines (47 loc) · 5.55 KB
Raw

AmbiQT and LogicalBeam

This repository contains the code and assets for the following paper, to appear at EMNLP 2023 (Main):

Benchmarking and Improving Text-to-SQL Generation Under Ambiguity
Adithya Bhaskar*, Tushar Tomar*, Ashutosh Sathe and Sunita Sarawagi

Figure Not Found

In this paper, we evaluate state-of-the-art and widely used models, including a fine-tuned T5-3B, ChatGPT and OpenAI Codex on converting English Questions to SQL queries under ambiguity. We show that all models fail to generate all plausible SQL queries in their top-k outputs, on the new benchmark we introduce, called AmbiQT. AmbiQT has over 3000 examples where one question has two distinct possible queries associated with it. We then propose a novel decoding scheme called LogicalBeam that when used with T5-3B, is 1.5-2.5X more effective at generating all plausible queries.

Please refer to our paper for more details.

Environment

Please install the latest versions of the libraries specified in requirements.txt, and those of PICARD if you wish to train your own model.

AmbiQT

The AmbiQT benchmark may be found under the benchmark/ directory. There is one subdirectory each for col-synonyms (Ambiguity of type "C" in the paper), tbl-synonyms ("T"), tbl-split ("J", Join Ambiguity) and tbl-agg ("P", Precomputed Aggregates). Each subdirectory is self-contained, and has

  • The train.json and validation.json files that we use for all of our experiments. We don't use or train on the former, but it may be helpful to others.
  • generate.py - This file can be used to generate the particular category of the benchmark given the original SPIDER files (you may have to modify the paths inside the file for this to work).
  • evaluate.py - This is the file used to evaluate the predictions of a particular method from a JSON file. Use the --file parameter to specify the input file, --pred_key to specify the key for the predictions in the JSON entries, and --k to specify k for selecting the top-k outputs.
  • For C and T types of ambiguities, we also provide a JSON file with the two synonyms generated with ChatGPT, that are used by generate.py to create the ambiguous queries.

Note: Please unzip db-content.zip to the directory db-content before running the evaluation scripts, or change DB_DIR in the corresponding evaluate.py file to point to database/ from SPIDER .

Outputs

The outputs generated by the various baselines, ablations and methods (including LogicalBeam) on AmbiQT are provided under the outputs/ directory.

Running LogicalBeam and its Ablations

Note: Please modify the REPO_ROOT variable in src/utils/globals.py to point to this repository, and unzip db-content.zip to the directory db-content before running the code.

We provide the source code of LogicalBeam in src/ directory. In particular,

  • src/utils contains helper functions to assist the rest of the code.
  • src/logicalbeam contains the main code of LogicalBeam, modularized into single tasks (Template Generation, Template Infilling, The Vanilla Text-to-SQL, and the Direct Text-to-SQL generation with plans).
  • src/main.py is the main entry point for LogicalBeam.
  • Template Infilling generates a list of lists of queries (a list for each template), totalling more than 5. The file src/collate.py collates the top-5 queries (as described in our paper) and produces an output where each entry's predictions are a single list of k entries.
  • src/chatgpt/ is self-contained and provides code for prompting ChatGPT for
    • Generating Synonyms for Columns (Used in constructing AmbiQT - C)
    • Generating Synonyms for Tables (Used in constructing AmbiQT - T), and
    • Evaluating ChatGPT on AmbiQT
  • The entry point into src/chatgpt/ is prompter.py. Note that, for all of the three modes above, if a request fails multiple times in a row, the current output is saved and the run is aborted. Be sure to pass in --resume to resume prompting from a partial output.

Checkpoints

We use/provide the following checkpoints for the various tasks.

Model Checkpoint
Baseline T5-3B (PICARD checkpoint) tscholak/cxmefzzi
Baseline Flan T5-XL adithyabhaskar/flan-t5-xl-t2s
Direct Ablation adithyabhaskar/flan-t5-xl-t2s-w-plan
Text-to-Template adithyabhaskar/flan-t5-xl-template-gen
Template-Fill-In adithyabhaskar/flan-t5-xl-template-fill

For the input format, see the README under training-files/.

Training your own version of the models

We used PICARD's trainer (with PICARD turned off) to train all of our models. We provide the datasets (one-to-one maps of SPIDER examples) under training-files/. We also detail the changes you need to make to PICARD to (a) train from your own file instead of the SPIDER splits, and (b) to introduce the new Template Match (a simple string comparison) metric to evaluate the Template Generation model in training-files/README.md. We left everything else untouched, i.e. the same as the default settings in PICARD. We also provide an example of the metric file(s) in training-files/metrics/.