338 research outputs found
Effective, efficient and reliable large language models
No full textIn recent years, Large Language Models (LLMs) have fundamentally transformed the field of Natural Language Processing (NLP), reshaping the landscape of AI research and applications. This thesis represents the culmination of four years of doctoral research, which began in 2020 when LLMs were still an emerging technology and GPT-3 had just been introduced. Over the course of this research, we have both observed and contributed to the advancement of some of the technologies underpinning LLMs, from their early stages to their current role as cutting-edge AI systems. Specifically, this thesis combines some of the works carried out during this time under three critical dimensions of LLMs: Effectiveness, Efficiency, and Reliability. On the Effectiveness dimension, we contributed to the development of instruction tuning - a key technique now ubiquitous in the training pipeline of LLMs. Our work demonstrated that smaller, instruction-tuned LLMs can outperform models up to 16 times their size, including GPT-3. We also developed PromptSource, an integrated development environment for creating, managing, and sharing natural language prompts, which has become a valuable resource for the NLP community. Both of these contributions were carried out during the BigScience Workshop, a year-long open research initiative by Hugging Face targeting the study of LLMs. Finally, along this dimension, we studied how to make these models handle multimodal database-like queries. Addressing the Efficiency dimension, we tackled the challenge of accelerating LLM inference. We introduced three novel parallel decoding algorithms that significantly speed up text generation without compromising output quality. This has since evolved into an active research area known as speculative or parallel decoding. Furthermore, we developed an efficient, language-specific instruction-tuned LLM for the Italian language, demonstrating a cost-effective approach to creating high-quality models for specific languages. Our research on Reliability addresses the critical issue of making these models reliable since they have been shown to systematically generate incorrect information - a phenomenon known as hallucinations. In this direction, we investigated whether it's possible to detect the model's confidence in its outputs. We conducted a comprehensive assessment of current uncertainty quantification methods and their evaluation protocols and explored novel approaches to combine these methods to improve the detection and quantification of uncertainty in LLM outputs. Our work paves the way for more Effective, Efficient, and Reliable large language models, addressing key challenges in their development and deployment while opening new avenues for future research in this rapidly evolving field
Aspirin, platelets, and cancer: The point of view of the internist.
No full textGrowing evidence suggests the beneficial effect of aspirin against some types of cancer, particularly of the gastrointestinal tract, and it has been provided for an effect both in cancer prevention as well as in survival improvement of cancer patients. Aspirin benefits increase with duration of treatment, especially after 10years of treatment. The inhibition of platelet activation at sites of gastrointestinal mucosal lesions could be the primary mechanism of action of low-dose aspirin. Indeed, the formation of tumor cell-induced platelet aggregates may favor immune evasion, by releasing angiogenic and growth factors, and also by promoting cancer cell dissemination. Moreover, platelets may contribute to aberrant COX-2 expression in colon carcinoma cells, thereby contributing to downregulation of oncosuppressor genes and upregulation of oncogenes, such as cyclin B1. Platelet adhesion to cancer cells leads also to an increased expression of genes involved in the EMT, such as the EMT-inducing transcription factors ZEB1 and TWIST1 and the mesenchymal marker vimentin. The aspirin-mediated inactivation of platelets may restore antitumor reactivity by blocking the release of paracrine lipid and protein mediators that induce COX-2 expression in adjacent nucleated cells at sites of mucosal injury. Thus, recent findings suggest interesting perspectives on "old" aspirin and NSAID treatment and/or "new" specific drugs to target the "evil" interactions between platelets and cancer for chemoprevention
Secure rendezvous and static containment in multi-agent systems with adversarial intruders
No full textIn this paper we propose a novel distributed local interaction protocol for networks of multi-agent systems (MASs) in a multi-dimensional space under directed time-varying graph with the objective to achieve secure rendezvous or static containment within the convex hull of a set of leader agents. We consider the scenario where a set of anonymous adversarial agents may intrude the network (or may be hijacked by a cyber-attack) and show that the proposed strategy guarantees the achievement of the global objective despite the continued influence of the adversaries which cannot be detected nor identified by the collaborative agents. We characterize the convergence properties of the proposed protocol in terms of the characteristics of the underlying network topology of the multi-agent system. Numerical simulations and examples corroborate the theoretical results
La semeiotica delmassiccio toracico anteriore rivisitata. Anterior chest wall examination reviewed.
No full textLa rassegna si propone di rivisitare lasemeiotica della parete toracica anteriore, evidenziando gli elementi clinici e strumentali utili nel guidare l'iter diagnostico
Coagulation at the crossroads of the communicable/non-communicable disease dyad: The case of pneumonia
No full textFauno: The Italian Large Language Model that will leave you senza parole!
No full textThis paper presents Fauno, the first and largest open-source Italian conversational Large Language Model
(LLM). Our goal with Fauno is to democratize the study of LLMs in Italian, demonstrating that obtaining
a fine-tuned conversational bot with a single GPU is possible. In addition, we release a collection of
datasets for conversational AI in Italian. The datasets on which we fine-tuned Fauno include various
topics such as general question answering, computer science, and medical questions. We release our
code and datasets on https://github.com/RSTLess-research/Fauno-Italian-LL
Significance of urinary 11-dehydro-thromboxane B2 in age-related diseases: Focus on atherothrombosis
No full textSyntNN at SemEval-2018 Task 2: is Syntax Useful for Emoji Prediction? Embedding Syntactic Trees in Multi Layer Perceptrons
No full textIn this paper, we present SyntNN as a way to include traditional syntactic models in multilayer neural networks used in the task of Semeval Task 2 of emoji prediction (Barbieri et al., 2018). The model builds on the distributed tree embedder also known as distributed tree kernel (Zanzotto and Dell’Arciprete, 2012). Initial results are extremely encouraging but additional analysis is needed to overcome the problem of overfitting
Comment on "Effect of extracorporeal shockwave therapy versus intra-articular injections of hyaluronic acid for the treatment of knee osteoarthritis"
No full textWe read with great interest the study by Lee et al. titled
“Effect of Extracorporeal Shockwave Therapy Versus
Intra-articular Injections of Hyaluronic Acid for the
Treatment of Knee Osteoarthritis. The authors were
able to evaluate and compare the effects and outcomes
of the extracorporeal shock wave therapy and intraarticular
injections of hyaluronic acid in patients with
knee osteoarthritis. This research is considered relevant
and interesting, especially because it is noted that the
authors compare an invasive therapy, such the intraarticular
injections, with the administration of an extracorporeal
show wave therapy, which is notably a kind
of therapy that does not involve the interruption of the
integrity of the skin. It is noted that the results from the
study are promising from a clinical aspec
- …
