SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.13315 · 2024

PuzzleVQA: Diagnosing Multimodal Reasoning Challenges of Language Models with Abstract Visual Patterns

arXiv · PDF · Open in the Atlas

Code that ran

14 of 14 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; 3 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 11 failures 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/2403.13315.json · badge: /api/ran/2403.13315.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
declare-lab/llm-puzzletest canonical (classifier label)
classifier basis: census
14 ran · 0 no recorded run
FunctionStatusWhere it lives
bfs_solve Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/checker_move.py
code served (permissive licence) · get_code("8915bd9c6d22d3f1")
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, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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.

# Setup code for bfs_solve function

# Analyzing the function:
# - start_state: a string that contains '_', 'G', 'R' characters
# - goal_state: a string with the same characters in a different arrangement
# - The function uses: state.index('_'), string indexing, string slicing, string concatenation
# - It performs BFS to find a path from start_state to goal_state

# Create simple valid puzzle states
# The puzzle appears to be a sliding puzzle variant with 'G' (green) and 'R' (red) pieces

start_state = "GG_RR"  # Starting configuration with empty space '_' in the middle
goal_state = "R_RGG"   # Goal configuration

ARGS = {
    "start_state": start_state,
    "goal_state": goal_state
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
choose_random_day Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/calendar_day.py
code served (permissive licence) · get_code("ddffea9a2c9f76aa")
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, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 random
import calendar

# Set seed for reproducibility
random.seed(42)

# Define arguments for choose_random_day
# The function takes year (int) and month (int) as parameters
ARGS = {
    "year": 2024,
    "month": 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
convert_image_to_bytes Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/data_loading.py
code served (permissive licence) · get_code("c305d316880e3c0e")
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, through the batch API) and returned bytes. 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.

from PIL import Image
import io

# Create a small test image
width, height = 100, 100
image = Image.new('RGB', (width, height), color='red')

ARGS = {
    'image': image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bytes — a bytes output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a bytes 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: none by the harness; any seed is in the setup block
convert_image_to_text Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/data_loading.py
code served (permissive licence) · get_code("e23bf02f45e9774d")
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, through the batch API) and returned str. 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 PIL import Image
import io

# Create a small PIL Image for testing
# The function expects a PIL Image object and calls:
# - image.save(output, format='PNG')
# This is a standard PIL Image method

# Create a simple 4x3 RGB image
image_array = np.random.randint(0, 256, size=(3, 4, 3), dtype=np.uint8)
test_image = Image.fromarray(image_array, mode='RGB')

ARGS = {
    'image': test_image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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: none by the harness; any seed is in the setup block
convert_text_to_image Ran
ran, honoured the contract we drafted
declare-lab/llm-puzzletest/AlgoPuzzleVQA/data_loading.py
code served (permissive licence) · get_code("d31d081e698b91b8")
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, through the batch API) and returned PngImageFile. 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 base64
import io
from PIL import Image

# Create a small test image
test_image = Image.new('RGB', (10, 10), color='red')

# Convert image to base64 string
image_buffer = io.BytesIO()
test_image.save(image_buffer, format='PNG')
image_buffer.seek(0)
image_base64 = base64.b64encode(image_buffer.read()).decode('utf-8')

ARGS = {
    'text': image_base64
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type PngImageFile — a PngImageFile output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a PngImageFile output the recorded ndim 3 and dtype family int 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: none by the harness; any seed is in the setup block
create_question Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/board_tile.py
code served (permissive licence) · get_code("459d2e0c665b8b99")
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, through the batch API) and returned str. 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.

# Set up valid arguments for create_question function

ARGS = {
    'rows': 8,
    'cols': 8,
    'num_removed': 1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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: none by the harness; any seed is in the setup block
day_of_week Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/calendar_day.py
code served (permissive licence) · get_code("e6a76ca3003b5abe")
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, through the batch API) and returned str. 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 calendar

# The function day_of_week takes three integer parameters: year, month, day
# It uses calendar.weekday() and calendar.day_name to return the day of week name

ARGS = {
    "year": 2024,
    "month": 1,
    "day": 15
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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: none by the harness; any seed is in the setup block
extract_program Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/python_executor.py
pointer only (licence: MIT); the stored copy get_code serves, from ChengpengLi1003/CoRT, is pointer only (licence: NONE) · get_code("1d961161c71cc153")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ChengpengLi1003/CoRT, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

generate_moves Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/checker_move.py
code served (permissive licence) · get_code("f78739cccda13e2d")
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, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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.

# Analyzing the function:
# - Takes a parameter 'state' which is used with:
#   - .index('_') method - needs a string or list with __contains__ and .index()
#   - indexing with [] and slicing
#   - .len() call
#   - equality comparisons with 'G' and 'R'
# - Returns a list of modified states
# The function appears to implement a puzzle game where '_' is empty,
# 'G' and 'R' are game pieces that can move in specific ways.

# The state parameter should be a string containing '_', 'G', and 'R' characters

ARGS = {
    "state": "G_R_GR"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
get_dimension_for_chain Ran
ran, honoured the contract we drafted
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/chain_link.py
code served (permissive licence) · get_code("fbca93db712d640f")
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, through the batch API) and returned tuple, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

torch.manual_seed(0)

# Analyzing the function:
# - total_chains_x: integer used in arithmetic
# - total_chains_y: integer used in arithmetic
# - rows: integer used in arithmetic
# - cols: integer used in arithmetic
# - width: integer with default 100, used in arithmetic
# - height: integer with default 60, used in arithmetic
# - offset_x: integer with default 50, used in arithmetic
# - offset_y: integer with default 50, used in arithmetic

ARGS = {
    "total_chains_x": 3,
    "total_chains_y": 4,
    "rows": 2,
    "cols": 5,
    "width": 100,
    "height": 60,
    "offset_x": 50,
    "offset_y": 50,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
get_xy_for_chain Ran
ran, honoured the contract we drafted
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/chain_link.py
code served (permissive licence) · get_code("123e9c69d0cccbea")
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, through the batch API) and returned tuple, ndim 2, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

torch.manual_seed(0)

ARGS = {
    "top_x": 10,
    "top_y": 20,
    "width": 100,
    "height": 60
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
month_calendar Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/calendar_day.py
code served (permissive licence) · get_code("170a11ba88de73b4")
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, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 calendar

# The function month_calendar takes two parameters:
# - year: an integer representing the year
# - month: an integer representing the month (1-12)

# Looking at the function body:
# - It calls calendar.month(year, month, 4, 2)
# - It performs string operations on the result
# - It expects year and month to be valid integers

ARGS = {
    "year": 2024,
    "month": 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
mutilated_chessboard Ran
ran; our contract draft was wrong, not the code
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/board_tile.py
code served (permissive licence) · get_code("df2734b5901e66c4")
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, through the batch API) 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
import matplotlib.pyplot as plt

fig, ax = plt.subplots(figsize=(6, 6))

ARGS = {
    "fig": fig,
    "ax": ax,
    "rows": 8,
    "cols": 8
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, 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
steps Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
declare-lab/llm-puzzletest/AlgoPuzzleVQA/generation/chain_link.py
code served (permissive licence) · get_code("836b31e7dc14c59e")
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, through the batch API) and returned dict. 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.

# Setup code for the steps function

# The function takes a list x and modifies it while tracking cuts and joins
# x[0] represents the number of joins accumulated
# x[1:] represents cuts remaining at each position
# The algorithm continues while x[0] < sum(x[1:])

# Create a simple list with initial values
# x[0] starts at 0 (joins), and x[1:] contains the cuts to be made
x = [0, 5, 3, 2]  # Start with 0 joins, and 5+3+2=10 total cuts to make

cut_time = 1  # Time per cut operation
join_time = 2  # Time per join operation

ARGS = {
    'x': x,
    'cut_time': cut_time,
    'join_time': join_time
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 13 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
extract_program 5 other papers
1 dated earlier, 4 later
2309.17452 (2023-09)
ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving
2406.18629 (2024-06)
Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs
2502.12067 (2025-02)
TokenSkip: Controllable Chain-of-Thought Compression in LLMs
and 2 more not named here

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 multimodal models extend the impressive capabilities of large language models by integrating multimodal understanding abilities. However, it is not clear how they can emulate the general intelligence and reasoning ability of humans. As recognizing patterns and abstracting concepts are key to general intelligence, we introduce PuzzleVQA, a collection of 2000 puzzle instances based on abstract patterns. With this dataset, we evaluate large multimodal models with abstract patterns based on fundamental concepts, including colors, numbers, sizes, and shapes. Through our experiments on state-of-the-art large multimodal models, we find that they are not able to generalize well to simple abstract patterns. Notably, GPT-4V achieves a score of 46.4% on single-concept puzzles, which shows that state-of-the-art models struggle on our dataset. To diagnose the reasoning challenges in large multimodal models, we progressively guide the models with our ground truth reasoning explanations for visual perception, inductive reasoning, and deductive reasoning. Our systematic analysis finds that the main bottlenecks of GPT-4V are weaker visual perception and inductive reasoning abilities. Through this work, we hope to shed light on the limitations of large multimodal models and how they can better emulate human cognitive processes in the future. Our data and code are available at https://puzzlevqa.github.io

For agents

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

get_harvested_code_for_paper("2403.13315")
get_code_for_paper("2403.13315")
have("2403.13315")

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

curl https://syntology.ai/api/ran/2403.13315.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.13315.svg)](https://syntology.ai/paper/2403.13315)

Connect an agent — have() is free.