Transformer-based models trained on text and vision modalities try to improve the performance on multimodal downstream tasks or tackle the problem Transformer-based models trained on text and vision modalities try to improve the performance on multimodal downstream tasks or tackle the problem of lack of grounding, e.g., addressing issues like models’ insufficient commonsense knowledge. While it is more straightforward to evaluate the effects of such models on multimodal tasks, such as visual question answering or image captioning, it is not as well-understood how these tasks affect the model itself, and its internal linguistic representations. In this work, we experiment with language models grounded in videos and measure the models’ performance on predicting masked words chosen based on their imageability. The results show that the smaller model benefits from video grounding in predicting highly imageable words, while the results for the larger model seem harder to interpret.of lack of grounding, e.g., addressing issues like models’ insufficient commonsense knowledge. While it is more straightforward to evaluate the effects of such models on multimodal tasks, such as visual question answering or image captioning, it is not as well-understood how these tasks affect the model itself, and its internal linguistic representations. In this work, we experiment with language models grounded in videos and measure the models’ performance on predicting masked words chosen based on their imageability. The results show that the smaller model benefits from video grounding in predicting highly imageable words, while the results for the larger model seem harder to interpret.

In recent years, instruction finetuning models have received increased attention due to their remarkable zero-shot and generalization capabilities. However, the widespread implementation of these models has been limited to the English language, largely due to the costs and challenges associated with creating instruction datasets. To overcome this, automatic instruction generation has been proposed as a resourceful alternative. We see this as an opportunity for the adoption of instruction finetuning for other languages. In this paper we explore the viability of instruction finetuning for Swedish. We translate a dataset of generated instructions from English to Swedish, using it to finetune both Swedish and non-Swedish models. Results indicate that the use of translated instructions significantly improves the models’ zero-shot performance, even on unseen data, while staying competitive with strong baselines ten times in size. We see this paper is a first step and a proof of concept that instruction finetuning for Swedish is within reach, through resourceful means, and that there exist several directions for further improvements.

Work on instruction-tuned Large Language Models (LLMs) has used automatic methods based on text overlap and LLM judgments as cost-effective alternatives to human evaluation. In this paper, we perform a meta-evaluation of such methods and assess their reliability across a broad range of tasks. In evaluating how well automatic methods align with human evaluations, correlation metrics are the most commonly employed method despite their inherent limitations when dealing with ties and different scales. To address these shortcomings, we use Pairwise Accuracy as an alternative to standard correlation measures. We observe that while automatic evaluation methods can approximate human ratings under specific conditions, their validity is highly context-dependent. Specifically, the simple ROUGE-L metric correlates very well with human ratings for short-answer English tasks but is unreliable in free-form generation tasks and cross-lingual scenarios. The effectiveness of the more advanced method of using GPT-4 as a judge diminishes significantly if reference answers are not included in the prompt, which is the scenario where this method has the potential to provide the most value compared to other metrics. Our findings enhance the understanding of how automatic methods should be applied and interpreted when developing and evaluating instruction-tuned LLMs.

Recent work on the Retrieval-Enhanced Transformer (Retro) model has shown that offloading memory from trainable weights to a retrieval database can significantly improve language modeling and match the performance of non-retrieval models that are an order of magnitude larger in size. It has been suggested that at least some of this performance gain is due to non-trivial generalization based on both model weights and retrieval. In this paper, we try to better understand the relative contributions of these two components. We find that the performance gains from retrieval largely originate from over-lapping tokens between the database and the test data, suggesting less non-trivial generalization than previously assumed. More generally, our results point to the challenges of evaluating the generalization of retrieval-augmented language models such as Retro, as even limited token overlap may significantly decrease test-time loss. We release our code and model at https://github.com/TobiasNorlund/retro

Augmenting language models with a retrieval mechanism has been shown to significantly improve their performance while keeping the number of parameters low. Retrieval-augmented models commonly rely on a semantic retrieval mechanism based on the similarity between dense representations of the query chunk and potential neighbors. In this paper, we study the state-of-the-art Retro model and observe that its performance gain is better explained by surface-level similarities, such as token overlap. Inspired by this, we replace the semantic retrieval in Retro with a surface-level method based on BM25, obtaining a significant reduction in perplexity. As full BM25 retrieval can be computationally costly for large datasets, we also apply it in a re-ranking scenario, gaining part of the perplexity reduction with minimal computational overhead.

Retrieval-augmented language models have demonstrated performance comparable to much larger models while requiring fewer computational resources. The effectiveness of these models crucially depends on the overlap between query and retrieved context, but the optimal degree of this overlap remains unexplored. In this paper, we systematically investigate how varying levels of query–context overlap affect model performance during both training and inference. Our experiments reveal that increased overlap initially has minimal effect, but substantially improves test-time perplexity and accelerates model learning above a critical threshold. Building on these findings, we demonstrate that deliberately increasing overlap through synthetic context can enhance data efficiency and reduce training time by approximately 40% without compromising performance. We specifically generate synthetic context through paraphrasing queries. We validate our perplexity-based findings on question-answering tasks, confirming that the benefits of retrieval-augmented language modeling extend to practical applications. Our results provide empirical evidence of significant optimization potential for retrieval mechanisms in language model pretraining.