More Like this

1. Exploring Human and Language Model Alignment in Perceived Design Similarity Using Ordinal Embeddings

   https://doi.org/10.1115/1.4069129  

   Keeler, Matthew; Fuge, Mark D; Peng, Aoran; Miller, Scarlett (October 2025, Journal of Mechanical Design)

   Abstract In recent years, large language models (LLMs) and vision language models (VLMs) have excelled at tasks requiring human-like reasoning, inspiring researchers in engineering design to use language models (LMs) as surrogate evaluators of design concepts. But do these models actually evaluate designs like humans? While recent work has shown that LM evaluations sometimes fall within human variance on Likert-scale grading tasks, those tasks often obscure the reasoning and biases behind the scores. To address this limitation, we compare LM word embeddings (trained to capture semantic similarity) with human-rated similarity embeddings derived from triplet comparisons (“is A closer to B than C?”) on a dataset of design sketches and descriptions. We assess alignment via local tripletwise similarity and embedding distances, allowing for deeper insights than raw Likert-scale scores provide. We also explore whether describing the designs to LMs through text or images improves alignment with human judgments. Our findings suggest that text alone may not fully capture the nuances humans key into, yet text-based embeddings outperform their multimodal counterparts on satisfying local triplets. On the basis of these insights, we offer recommendations for effectively integrating LMs into design evaluation tasks. 

   more » « less
2. Empowering Language Models with Knowledge Graph Reasoning for Question Answering

   Hu, Ziniu; Xu, Yichong; Yu, Wenhao; Wang, Shuohang; Yang, Ziyi; Zhu, Chenguang; Chang, Kai-Wei; Sun, Yizhou (December 2022, Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing)

   Answering open-domain questions requires world knowledge about in-context entities. As pre-trained Language Models (LMs) lack the power to store all required knowledge, external knowledge sources, such as knowledge graphs, are often used to augment LMs. In this work, we propose knOwledge REasOning empowered Language Model (OREOLM), which consists of a novel Knowledge Interaction Layer that can be flexibly plugged into existing Transformer-based LMs to interact with a differentiable Knowledge Graph Reasoning module collaboratively. In this way, LM guides KG to walk towards the desired answer, while the retrieved knowledge improves LM. By adopting OREOLM to RoBERTa and T5, we show significant performance gain, achieving state-of-art results in the Closed-Book setting. The performance enhancement is mainly from the KG reasoning’s capacity to infer missing relational facts. In addition, OREOLM provides reasoning paths as rationales to interpret the model’s decision. 

   more » « less
3. Can You Learn Semantics Through Next-Word Prediction? The Case of Entailment

   Merrill, William; Wu, Zhaofeng; Naka, Norihito; Kim, Yoon; Linzen, Tal (August 2024, Annual Meeting of the Association for Computational Linguistics 2024)

   Do LMs infer the semantics of text from co-occurrence patterns in their training data? Merrill et al. (2022) argue that, in theory, sentence co-occurrence probabilities predicted by an optimal LM should reflect the entailment relationship of the constituent sentences, but it is unclear whether probabilities predicted by neural LMs encode entailment in this way because of strong assumptions made by Merrill et al. (namely, that humans always avoid redundancy). In this work, we investigate whether their theory can be used to decode entailment relations from neural LMs. We find that a test similar to theirs can decode entailment relations between natural sentences, well above random chance, though not perfectly, across many datasets and LMs. This suggests LMs implicitly model aspects of semantics to predict semantic effects on sentence co-occurrence patterns. However, we find the test that predicts entailment in practice works in the opposite direction to the theoretical test. We thus revisit the assumptions underlying the original test, finding its derivation did not adequately account for redundancy in human-written text. We argue that better accounting for redundancy related to explanations might derive the observed flipped test and, more generally, improve computational models of speakers in linguistics. 

   more » « less
4. LitCab: Lightweight Language Model Calibration over Short- and Long-form Responses

   Liu, Xin; Khalifa, Muhammad; Wang, Lu (May 2024, Proceedings of the International Conference on Learning Representations (ICLR))

   A model is considered well-calibrated when its probability estimate aligns with the actual likelihood of the output being correct. Calibrating language models (LMs) is crucial, as it plays a vital role in detecting and mitigating hallucinations of LMs as well as building more trustworthy models. However, standard calibration techniques may not be suited for LM calibration. For instance, post-processing methods such as temperature scaling do not reorder the candidate generations. On the other hand, training-based methods require fine-tuning the entire model, which is impractical for LMs of large scale. We present LITCAB, a lightweight calibration mechanism consisting of a single linear layer that takes the input text representation and predicts a bias term, which is then added to the LM output logits. LITCAB improves model calibration by only adding < 2% of the original model parameters. For evaluation, we construct CAT, a benchmark consisting of eight text generation tasks, covering responses ranging from short phrases to paragraphs. We test LITCAB with Llama2-7B, where it improves calibration across all tasks, reducing the average ECE score by as large as 30%. We further conduct a comprehensive evaluation with multiple popular open-sourced LMs from GPT and LLaMA families, yielding the following key findings: (i) Larger models within the same family exhibit better calibration on tasks with short generation tasks, but not necessarily for longer ones. (ii) GPT-family models show superior calibration compared to LLaMA, Llama2, and Vicuna models, despite having much fewer parameters. (iii) Fine-tuning pretrained model (e.g., LLaMA) with samples of limited purpose (e.g., conversations) may lead to worse calibration, highlighting the importance of fine-tuning setups for calibrating LMs. 

   more » « less
5. From music to animacy: Causal reasoning links animate agents with musical sounds

   Kim, M.; Schachner, A. (July 2021, Proceedings of the Annual Conference of the Cognitive Science Society)

   Fitch, T.; Lamm, C.; Leder, H.; Teßmar-Raible, K. (Ed.)

   Listening to music activates representations of movement and social agents. Why? We ask whether high-level causal reasoning about how music was generated can lead people to link musical sounds with animate agents. To test this, we asked whether people (N=60) make flexible inferences about whether an agent caused musical sounds, integrating information from the sounds’ timing and from the visual context in which it was produced. Using a 2x2 within-subject design, we found evidence of causal reasoning: In a context where producing a musical sequence would require self-propelled movement, people inferred that an agent had been present causing the sounds. When the context provided an alternative possible explanation, this ‘explained away’ the agent, reducing the tendency to infer an agent was present for the same acoustic stimuli. People can use causal reasoning to infer whether an agent produced musical sounds, suggesting that high-level cognition can link music with social concepts. 

   more » « less