SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2010.11246 · EMNLP Findings · 2020

On the Potential of Lexico-logical Alignments for Semantic Parsing to SQL Queries

Hal Daumé, Jordan Boyd-Graber, Chen Zhao, Tianze Shi, Lillian Lee

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 1 failure of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2010.11246.json · badge: /api/ran/2010.11246.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
microsoft/Table-Pretraining community
Papers with Code archive
3 ran · 0 no recorded run
tzshi/squall marked official in Papers with Code
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
RowDeleteTruncate Ran
ran on a synthesized input
microsoft/Table-Pretraining/tapex/processor/table_truncate.py
code served (permissive licence) · get_code("d816de85317c8f3e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned NoneType. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from transformers import AutoTokenizer, BasicTokenizer

# Create a simple mock TableLinearize implementation
class MockTableLinearize:
    def process_table(self, table_content):
        header_str = self.process_header(table_content['header'])
        rows_str = ' '.join([self.process_row(row, i+1) for i, row in enumerate(table_content['rows'])])
        return header_str + ' ' + rows_str
    
    def process_header(self, headers):
        return ' | '.join(headers)
    
    def process_row(self, row, row_index):
        return f"row {row_index}: " + ' | '.join(row)

# Create a simple mock tokenizer
class MockTokenizer:
    def tokenize(self, text, add_special_tokens=False):
        # Simple whitespace tokenization
        tokens = text.split()
        if add_special_tokens:
            tokens = ['<s>'] + tokens + ['</s>']
        return tokens
    
    def __call__(self, text, **kwargs):
        return {'input_ids': list(range(len(self.tokenize(text))))}

# Initialize arguments for RowDeleteTruncate.__init__
table_linearize = MockTableLinearize()
tokenizer = MockTokenizer()
max_input_length = 50

INIT_ARGS = {
    'table_linearize': table_linearize,
    'tokenizer': tokenizer,
    'max_input_length': max_input_length
}

# Create sample table content for the truncate_table method
table_content = {
    'header': ['Name', 'Age', 'City'],
    'rows': [
        ['Alice', '25', 'New York'],
        ['Bob', '30', 'Los Angeles'],
        ['Charlie', '35', 'Chicago'],
        ['David', '40', 'Houston'],
        ['Eve', '45', 'Phoenix']
    ]
}

question = "What is the age of Alice?"
answer = ['25', 'Alice']

ARGS = {
    'table_content': table_content,
    'question': question,
    'answer': answer
}

METHOD = 'truncate_table'
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NoneType — a NoneType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoneType output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
TableLinearize Ran
ran on a synthesized input
microsoft/Table-Pretraining/tapex/processor/table_truncate.py
code served (permissive licence) · get_code("b1309c93486f799f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned NoneType. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import abc
from typing import List, Dict

class TableLinearize(abc.ABC):
    PROMPT_MESSAGE = '\n        Please check that your table must follow the following format:\n        {"header": ["col1", "col2", "col3"], "rows": [["row11", "row12", "row13"], ["row21", "row22", "row23"]]}\n    '

    def process_table(self, table_content: Dict) -> str:
        """
        Given a table, TableLinearize aims at converting it into a flatten sequence with special symbols.
        """
        pass

    def process_header(self, headers: List):
        """
        Given a list of headers, TableLinearize aims at converting it into a flatten sequence with special symbols.
        """
        pass

    def process_row(self, row: List, row_index: int):
        """
        Given a row, TableLinearize aims at converting it into a flatten sequence with special symbols.
        """
        pass


class ConcreteTableLinearize(TableLinearize):
    def __init__(self):
        pass
    
    def process_table(self, table_content: Dict) -> str:
        result = []
        if "header" in table_content:
            result.append(self.process_header(table_content["header"]))
        if "rows" in table_content:
            for idx, row in enumerate(table_content["rows"]):
                result.append(self.process_row(row, idx))
        return " ".join(result)
    
    def process_header(self, headers: List):
        return "| " + " | ".join(str(h) for h in headers) + " |"
    
    def process_row(self, row: List, row_index: int):
        return f"row{row_index}: | " + " | ".join(str(cell) for cell in row) + " |"


INIT_ARGS = {}

ARGS = {
    "table_content": {
        "header": ["Name", "Age", "City"],
        "rows": [
            ["Alice", "25", "NYC"],
            ["Bob", "30", "LA"]
        ]
    }
}

METHOD = "process_table"

instance = ConcreteTableLinearize(**INIT_ARGS)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NoneType — a NoneType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoneType output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
TableTruncate Ran
ran on a synthesized input
microsoft/Table-Pretraining/tapex/processor/table_truncate.py
code served (permissive licence) · get_code("d49b92742c05beb7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned NoneType. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from transformers import AutoTokenizer, PreTrainedTokenizerFast
from tokenizers import Tokenizer, models, pre_tokenizers, trainers

# Create a minimal mock tokenizer instead of downloading
def create_mock_tokenizer():
    # Create a simple BPE tokenizer from scratch
    tokenizer = Tokenizer(models.BPE())
    tokenizer.pre_tokenizer = pre_tokenizers.Whitespace()
    
    # Train on a small vocabulary
    trainer = trainers.BpeTrainer(vocab_size=100, special_tokens=["<s>", "</s>", "<pad>", "<unk>"])
    tokenizer.train_from_iterator([
        "question answer table row column header",
        "what is the value in the table",
        "how many rows are there"
    ], trainer)
    
    # Wrap in PreTrainedTokenizerFast
    wrapped = PreTrainedTokenizerFast(tokenizer_object=tokenizer)
    wrapped.pad_token = "<pad>"
    wrapped.eos_token = "</s>"
    wrapped.bos_token = "<s>"
    wrapped.unk_token = "<unk>"
    
    return wrapped

mock_tokenizer = create_mock_tokenizer()

INIT_ARGS = {
    'tokenizer': mock_tokenizer,
    'max_input_length': 512
}

table_content = {
    'header': ['Name', 'Age', 'City', 'Score'],
    'rows': [
        ['Alice', '25', 'NYC', '95'],
        ['Bob', '30', 'LA', '87'],
        ['Charlie', '35', 'Chicago', '92']
    ],
    'title': 'Sample Table',
    'caption': 'This is a sample table with data'
}

question = "What is the score of Alice?"

answer = ['95', 'Alice has a score of 95']

ARGS = {
    'table_content': table_content,
    'question': question,
    'answer': answer
}

METHOD = 'truncate_table'
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, tokenizers, torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NoneType — a NoneType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoneType output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
attention_matrix_col Ran
ran; our contract draft was wrong, not the code
tzshi/squall/model/utils.py
code served (permissive licence) · get_code("df5ba2f0e76b3eaf")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Construct a valid instance dict for the attention_matrix_col function
instance = {
    'nl': ['what', 'is', 'the', 'name'],  # 4 natural language tokens
    'columns': ['c1', 'c2', 'c3'],  # 3 columns
    'sql': [
        ('Column', 'c1_0', (0, 1)),
        ('Other', 'some_value', (1, 2)),
        ('Column', 'c2_0', (2, 3)),
    ],
    'align': [
        ([0, 1], [0]),  # nl positions [0, 1] align with sql step 0
        ([2], [2]),     # nl position [2] aligns with sql step 2
        ([3], [2]),     # nl position [3] aligns with sql step 2
    ]
}

ARGS = {
    'instance': instance
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
TableLinearize
brick of this paper's assembly: helper
1 other paper
1 later
2305.14303 (2023-05)
QTSUMM: Query-Focused Summarization over Tabular Data

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Large-scale semantic parsing datasets annotated with logical forms have enabled major advances in supervised approaches. But can richer supervision help even more? To explore the utility of fine-grained, lexical-level supervision, we introduce SQUALL, a dataset that enriches 11,276 WIKITABLEQUESTIONS English-language questions with manually created SQL equivalents plus alignments between SQL and question fragments. Our annotation enables new training possibilities for encoderdecoder models, including approaches from machine translation previously precluded by the absence of alignments. We propose and test two methods: (1) supervised attention; (2) adopting an auxiliary objective of disambiguating references in the input queries to table columns. In 5-fold cross validation, these strategies improve over strong baselines by 4.4% execution accuracy. Oracle experiments suggest that annotated alignments can support further accuracy gains of up to 23.9%.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2010.11246")
get_code_for_paper("2010.11246")
have("2010.11246")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2010.11246.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2010.11246.svg)](https://syntology.ai/paper/2010.11246)

Connect an agent — have() is free.