REVIEW 8 cited by
Understanding Alignment in Multimodal LLMs: A Comprehensive Study
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
Signed reviews
read the original abstract
Preference alignment has become a crucial component in enhancing the performance of Large Language Models (LLMs), yet its impact in Multimodal Large Language Models (MLLMs) remains comparatively underexplored. Similar to language models, MLLMs for image understanding tasks encounter challenges like hallucination. In MLLMs, hallucination can occur not only by stating incorrect facts but also by producing responses that are inconsistent with the image content. A primary objective of alignment for MLLMs is to encourage these models to align responses more closely with image information. Recently, multiple works have introduced preference datasets for MLLMs and examined different alignment methods, including Direct Preference Optimization (DPO) and Proximal Policy Optimization (PPO). However, due to variations in datasets, base model types, and alignment methods, it remains unclear which specific elements contribute most significantly to the reported improvements in these works. In this paper, we independently analyze each aspect of preference alignment in MLLMs. We start by categorizing the alignment algorithms into two groups, offline (such as DPO), and online (such as online-DPO), and show that combining offline and online methods can improve the performance of the model in certain scenarios. We review a variety of published multimodal preference datasets and discuss how the details of their construction impact model performance. Based on these insights, we introduce a novel way of creating multimodal preference data called Bias-Driven Hallucination Sampling (BDHS) that needs neither additional annotation nor external models, and show that it can achieve competitive performance to previously published alignment work for multimodal models across a range of benchmarks.
Forward citations
Cited by 8 Pith papers
-
Toward a Theory of Value in AI Alignment
A systematic annotation of 94 AI alignment papers shows the field largely equates human values with measurable preferences, rarely defines values, and is increasingly removing humans from alignment evaluation.
-
Groc-PO: Grounded Context Preference Optimization for Truthful Multimodal LLMs
Groc-PO applies preference optimization at three grounded stages — object grounding, contextual grounding, grounded reasoning — and outperforms final-answer-only DPO on hallucination and complex-reasoning benchmarks.
-
Controlling Multimodal LLMs via Reward-guided Decoding
MRGD guides MLLM decoding with a learned hallucination reward and a detector-based recall reward, allowing users to trade off object precision, recall, and test-time compute while reducing object hallucinations on CHA...
-
TUNA: Comprehensive Fine-grained Temporal Understanding Evaluation on Dense Dynamic Videos
TUNA introduces a 1,000-video benchmark with dense temporal captions and 1,432 multiple-choice questions, and finds that current video LMMs are weakest at camera motion, action sequences, and multi-subject scenes.
-
Beyond Human Data: Aligning Multimodal Large Language Models by Iterative Self-Evolution
A multimodal LLM can improve itself using only unlabeled images by self-generating questions, self-enhancing answers, and adding a description-alignment loss to DPO.
-
PerPO: Perceptual Preference Optimization via Discriminative Rewarding
PerPO trains multimodal LLMs by ranking their candidate answers with deterministic visual rewards (IoU, edit distance) and using the reward differences as margins in listwise preference optimization.
-
Efficient Self-Improvement in Multimodal Large Language Models: A Model-Level Judge-Free Approach
A judge-free self-improvement pipeline for multimodal LLMs that generates hallucinated caption pairs with a controlled decoding ratio, filters and swaps them with CLIP scores, and trains with DPO, reporting reduced ha...
-
Align Anything: Training All-Modality Models to Follow Instructions with Language Feedback
The paper proposes learning from language feedback to synthesize multimodal preference pairs, but the evidence is weakened by an undefined improvement metric and small, unvalidated effect sizes.
Discussion (0). Continue with ORCID to comment.