Pith. sign in

REVIEW 5 cited by

Can Large Language Models Understand Symbolic Graphics Programs?

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2408.08313 v4 pith:35JUSVIL submitted 2024-08-15 cs.LG cs.AIcs.CLcs.CV

Can Large Language Models Understand Symbolic Graphics Programs?

classification cs.LG cs.AIcs.CLcs.CV
keywords programssymbolicgraphicsllmsabilitylargemodelsreasoning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Against the backdrop of enthusiasm for large language models (LLMs), there is a growing need to scientifically assess their capabilities and shortcomings. This is nontrivial in part because it is difficult to find tasks which the models have not encountered during training. Utilizing symbolic graphics programs, we propose a domain well-suited to test multiple spatial-semantic reasoning skills of LLMs. Popular in computer graphics, these programs procedurally generate visual data. While LLMs exhibit impressive skills in general program synthesis and analysis, symbolic graphics programs offer a new layer of evaluation: they allow us to test an LLM's ability to answer semantic questions about the images or 3D geometries without a vision encoder. To semantically understand the symbolic programs, LLMs would need to possess the ability to "imagine" and reason how the corresponding graphics content would look with only the symbolic description of the local curvatures and strokes. We use this task to evaluate LLMs by creating a large benchmark for the semantic visual understanding of symbolic graphics programs, built procedurally with minimal human effort. Particular emphasis is placed on transformations of images that leave the image level semantics invariant while introducing significant changes to the underlying program. We evaluate commercial and open-source LLMs on our benchmark to assess their ability to reason about visual output of programs, finding that LLMs considered stronger at reasoning generally perform better. Lastly, we introduce a novel method to improve this ability -- Symbolic Instruction Tuning (SIT), in which the LLM is finetuned with pre-collected instruction data on symbolic graphics programs. Interestingly, we find that SIT not only improves LLM's understanding on symbolic programs, but it also improves general reasoning ability on various other benchmarks.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. DisciplineGen-1M: A Large-Scale Dataset for Multidisciplinary Visual Generation and Editing

    cs.CV 2026-07 unverdicted novelty 7.0

    DisciplineGen-1M is a million-scale multidisciplinary dataset for text-to-image generation and editing, paired with a discipline-informed model that improves results on discipline-specific benchmarks.

  2. mEOL: Training-Free Instruction-Guided Multimodal Embedder for Vector Graphics and Image Retrieval

    cs.CV 2026-04 unverdicted novelty 7.0

    mEOL creates aligned embeddings for text, images, and SVGs using instruction-guided MLLM one-word summaries and semantic SVG rewriting, outperforming baselines on a new text-to-SVG retrieval benchmark.

  3. CodePercept: Code-Grounded Visual STEM Perception for MLLMs

    cs.CV 2026-03 conditional novelty 6.5

    Perception, not reasoning, is the main bottleneck for MLLM STEM visual reasoning, and training on executable reconstruction code measurably fixes it.

  4. BrepCoder: A Unified Multimodal Large Language Model for Multi-task B-rep Reasoning

    cs.LG 2026-02 conditional novelty 6.0

    BrepCoder trains one multimodal LLM to read B-rep CAD geometry and output CAD code, handling reverse engineering, completion, error correction, and CAD-QA.

  5. InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency

    cs.CV 2025-08 unverdicted novelty 6.0

    InternVL3.5 advances open-source multimodal models with Cascade RL for +16% reasoning gains and ViR for 4x inference speedup, with the 241B model reaching SOTA among open-source MLLMs on multimodal, reasoning, and age...