In an era where visual content generation is increasingly driven by machine learning, the integration of human feedback into generative models presents significant opportunities for enhancing user experience and output quality. This study explores strategies for incorporating iterative human feedback into the generative process of diffusion-based text-to-image models. We propose FABRIC, a training-free approach applicable to a wide range of popular diffusion models, which exploits the self-attention layer present in the most widely used architectures to condition the diffusion process on a set of feedback images. To ensure a rigorous assessment of our approach, we introduce a comprehensive evaluation methodology, offering a robust mechanism to quantify the performance of generative visual models that integrate human feedback. We show that generation results improve over multiple rounds of iterative feedback through exhaustive analysis, implicitly optimizing arbitrary user preferences. The potential applications of these findings extend to fields such as personalized content creation and customization.
We describe a compression-aware method to compute all-vs-all maximal exact matches (MEM) among strings of a repetitive collection $\mathcal{T}$. The key concept in our work is the construction of a fully-balanced grammar $\mathcal{G}$ from $\mathcal{T}$ that meets a property that we call \emph{fix-free}: the expansions of the nonterminals that have the same height in the parse tree form a fix-free set (i.e., prefix-free and suffix-free). The fix-free property allows us to compute the MEMs of $\mathcal{T}$ incrementally over $\mathcal{G}$ using a standard suffix-tree-based MEM algorithm, which runs on a subset of grammar rules at a time and does not decompress nonterminals. By modifying the locally-consistent grammar of Christiansen et al 2020., we show how we can build $\mathcal{G}$ from $\mathcal{T}$ in linear time and space. We also demonstrate that our MEM algorithm runs on top of $\mathcal{G}$ in $O(G +occ)$ time and uses $O(\log G(G+occ))$ bits, where $G$ is the grammar size, and $occ$ is the number of MEMs in $\mathcal{T}$. In the conclusions, we discuss how our idea can be modified to implement approximate pattern matching in compressed space.
Electrocardiogram (ECG) synthesis is the area of research focused on generating realistic synthetic ECG signals for medical use without concerns over annotation costs or clinical data privacy restrictions. Traditional ECG generation models consider a single ECG lead and utilize GAN-based generative models. These models can only generate single lead samples and require separate training for each diagnosis class. The diagnosis classes of ECGs are insufficient to capture the intricate differences between ECGs depending on various features (e.g. patient demographic details, co-existing diagnosis classes, etc.). To alleviate these challenges, we present a text-to-ECG task, in which textual inputs are used to produce ECG outputs. Then we propose Auto-TTE, an autoregressive generative model conditioned on clinical text reports to synthesize 12-lead ECGs, for the first time to our knowledge. We compare the performance of our model with other representative models in text-to-speech and text-to-image. Experimental results show the superiority of our model in various quantitative evaluations and qualitative analysis. Finally, we conduct a user study with three board-certified cardiologists to confirm the fidelity and semantic alignment of generated samples. our code will be available at https://github.com/TClife/text_to_ecg
Modular vision-language models (Vision-LLMs) align pretrained image encoders with (pretrained) large language models (LLMs), representing a computationally much more efficient alternative to end-to-end training of large vision-language models from scratch, which is prohibitively expensive for most. Vision-LLMs instead post-hoc condition LLMs to `understand' the output of an image encoder. With the abundance of readily available high-quality English image-text data as well as monolingual English LLMs, the research focus has been on English-only Vision-LLMs. Multilingual vision-language models are still predominantly obtained via expensive end-to-end pretraining, resulting in comparatively smaller models, trained on limited multilingual image data supplemented with text-only multilingual corpora. In this work, we present mBLIP, the first multilingual Vision-LLM, which we obtain in a computationally efficient manner -- on consumer hardware using only a few million training examples -- by leveraging a pretrained multilingual LLM. To this end, we \textit{re-align} an image encoder previously tuned to an English LLM to a new, multilingual LLM -- for this, we leverage multilingual data from a mix of vision-and-language tasks, which we obtain by machine-translating high-quality English data to 95 languages. On the IGLUE benchmark, mBLIP yields results competitive with state-of-the-art models. Moreover, in image captioning on XM3600, mBLIP (zero-shot) even outperforms PaLI-X (a model with 55B parameters). Compared to these very large multilingual vision-language models trained from scratch, we obtain mBLIP by training orders of magnitude fewer parameters on magnitudes less data. We release our model and code at \url{https://github.com/gregor-ge/mBLIP}.
Language Models and Vision Language Models have recently demonstrated unprecedented capabilities in terms of understanding human intentions, reasoning, scene understanding, and planning-like behaviour, in text form, among many others. In this work, we investigate how to embed and leverage such abilities in Reinforcement Learning (RL) agents. We design a framework that uses language as the core reasoning tool, exploring how this enables an agent to tackle a series of fundamental RL challenges, such as efficient exploration, reusing experience data, scheduling skills, and learning from observations, which traditionally require separate, vertically designed algorithms. We test our method on a sparse-reward simulated robotic manipulation environment, where a robot needs to stack a set of objects. We demonstrate substantial performance improvements over baselines in exploration efficiency and ability to reuse data from offline datasets, and illustrate how to reuse learned skills to solve novel tasks or imitate videos of human experts.
Despite previous efforts in melody-to-lyric generation research, there is still a significant compatibility gap between generated lyrics and melodies, negatively impacting the singability of the outputs. This paper bridges the singability gap with a novel approach to generating singable lyrics by jointly Learning wOrding And Formatting during Melody-to-Lyric training (LOAF-M2L). After general-domain pretraining, our proposed model acquires length awareness first from a large text-only lyric corpus. Then, we introduce a new objective informed by musicological research on the relationship between melody and lyrics during melody-to-lyric training, which enables the model to learn the fine-grained format requirements of the melody. Our model achieves 3.75% and 21.44% absolute accuracy gains in the outputs' number-of-line and syllable-per-line requirements compared to naive fine-tuning, without sacrificing text fluency. Furthermore, our model demonstrates a 63.92% and 74.18% relative improvement of music-lyric compatibility and overall quality in the subjective evaluation, compared to the state-of-the-art melody-to-lyric generation model, highlighting the significance of formatting learning.
Text summarization is an essential task in natural language processing, and researchers have developed various approaches over the years, ranging from rule-based systems to neural networks. However, there is no single model or approach that performs well on every type of text. We propose a system that recommends the most suitable summarization model for a given text. The proposed system employs a fully connected neural network that analyzes the input content and predicts which summarizer should score the best in terms of ROUGE score for a given input. The meta-model selects among four different summarization models, developed for the Slovene language, using different properties of the input, in particular its Doc2Vec document representation. The four Slovene summarization models deal with different challenges associated with text summarization in a less-resourced language. We evaluate the proposed SloMetaSum model performance automatically and parts of it manually. The results show that the system successfully automates the step of manually selecting the best model.
We study the problem of decomposing a complex text-to-sql task into smaller sub-tasks and how such a decomposition can significantly improve the performance of Large Language Models (LLMs) in the reasoning process. There is currently a significant gap between the performance of fine-tuned models and prompting approaches using LLMs on challenging text-to-sql datasets such as Spider. We show that SQL queries, despite their declarative structure, can be broken down into sub-problems and the solutions of those sub-problems can be fed into LLMs to significantly improve their performance. Our experiments with three LLMs show that this approach consistently improves their performance by roughly 10%, pushing the accuracy of LLMs towards state-of-the-art, and even beating large fine-tuned models on the holdout Spider dataset.
Vector representations of natural language are ubiquitous in search applications. Recently, various methods based on contrastive learning have been proposed to learn textual representations from unlabelled data; by maximizing alignment between minimally-perturbed embeddings of the same text, and encouraging a uniform distribution of embeddings across a broader corpus. Differently, we propose maximizing alignment between texts and a composition of their phrasal constituents. We consider several realizations of this objective and elaborate the impact on representations in each case. Experimental results on semantic textual similarity tasks show improvements over baselines that are comparable with state-of-the-art approaches. Moreover, this work is the first to do so without incurring costs in auxiliary training objectives or additional network parameters.
Long sequences of text are challenging in the context of transformers, due to quadratic memory increase in the self-attention mechanism. As this issue directly affects the translation from natural language to SQL queries (as techniques usually take as input a concatenated text with the question and the database schema), we present techniques that allow long text sequences to be handled by transformers with up to 512 input tokens. We propose a training process with database schema pruning (removal of tables and columns names that are useless for the query of interest). In addition, we used a multilingual approach with the mT5-large model fine-tuned with a data-augmented Spider dataset in four languages simultaneously: English, Portuguese, Spanish, and French. Our proposed technique used the Spider dataset and increased the exact set match accuracy results from 0.718 to 0.736 in a validation dataset (Dev). Source code, evaluations, and checkpoints are available at: \underline{https://github.com/C4AI/gap-text2sql}.