Existing methods for adapting large language models (LLMs) to new tasks are not suited to multi-task adaptation because they modify all the model weights -- causing destructive interference between tasks. The resulting effects, such as catastrophic forgetting of earlier tasks, make it challenging to obtain good performance on multiple tasks at the same time. To mitigate this, we propose Lottery Ticket Adaptation (LoTA), a sparse adaptation method that identifies and optimizes only a sparse subnetwork of the model. We evaluate LoTA on a wide range of challenging tasks such as instruction following, reasoning, math, and summarization. LoTA obtains better performance than full fine-tuning and low-rank adaptation (LoRA), and maintains good performance even after training on other tasks -- thus, avoiding catastrophic forgetting. By extracting and fine-tuning over lottery tickets (or sparse task vectors), LoTA also enables model merging over highly dissimilar tasks. Our code is made publicly available at https://github.com/kiddyboots216/lottery-ticket-adaptation.

As large language models gain widespread adoption, running them efficiently becomes crucial. Recent works on LLM inference use speculative decoding to achieve extreme speedups. However, most of these works implicitly design their algorithms for high-end datacenter hardware. In this work, we ask the opposite question: how fast can we run LLMs on consumer machines? Consumer GPUs can no longer fit the largest available models (50B+ parameters) and must offload them to RAM or SSD. When running with offloaded parameters, the inference engine can process batches of hundreds or thousands of tokens at the same time as just one token, making it a natural fit for speculative decoding. We propose SpecExec (Speculative Execution), a simple parallel decoding method that can generate up to 20 tokens per target model iteration for popular LLM families. It utilizes the high spikiness of the token probabilities distribution in modern LLMs and a high degree of alignment between model output probabilities. SpecExec takes the most probable tokens continuation from the draft model to build a "cache" tree for the target model, which then gets validated in a single pass. Using SpecExec, we demonstrate inference of 50B+ parameter LLMs on consumer GPUs with RAM offloading at 4-6 tokens per second with 4-bit quantization or 2-3 tokens per second with 16-bit weights.

Large language models (LLMs) have fundamentally transformed artificial intelligence, catalyzing recent advancements while imposing substantial environmental and computational burdens. We introduce TRAWL (Tensor Reduced and Approximated Weights for Large Language Models), a novel methodology for optimizing LLMs through tensor decomposition. TRAWL leverages diverse strategies to exploit matrices within transformer-based architectures, realizing notable performance enhancements without necessitating retraining. The most significant improvements were observed through a layer-by-layer intervention strategy, particularly when applied to fully connected weights of the final layers, yielding up to 16% enhancement in accuracy without the need for additional data or fine-tuning. These results underscore the importance of targeted and adaptive techniques in increasing the efficiency and effectiveness of large language model optimization, thereby promoting the development of more sustainable and accessible AI systems.

In-context learning (ICL) empowers large language models (LLMs) to perform diverse tasks in underrepresented languages using only short in-context information, offering a crucial avenue for narrowing the gap between high-resource and low-resource languages. Nonetheless, there is only a handful of works explored ICL for low-resource languages with most of them focusing on relatively high-resource languages, such as French and Spanish. In this work, we extensively study ICL and its cross-lingual variation (X-ICL) on 25 low-resource and 7 relatively higher-resource languages. Our study not only assesses the effectiveness of ICL with LLMs in low-resource languages but also identifies the shortcomings of in-context label alignment, and introduces a more effective alternative: query alignment. Moreover, we provide valuable insights into various facets of ICL for low-resource languages. Our study concludes the significance of few-shot in-context information on enhancing the low-resource understanding quality of LLMs through semantically relevant information by closing the language gap in the target language and aligning the semantics between the targeted low-resource and the high-resource language that the model is proficient in. Our work highlights the importance of advancing ICL research, particularly for low-resource languages. Our code is publicly released at https://github.com/SamuelCahyawijaya/in-context-alignment

This paper presents a systematic defense of large language model (LLM) hallucinations or 'confabulations' as a potential resource instead of a categorically negative pitfall. The standard view is that confabulations are inherently problematic and AI research should eliminate this flaw. In this paper, we argue and empirically demonstrate that measurable semantic characteristics of LLM confabulations mirror a human propensity to utilize increased narrativity as a cognitive resource for sense-making and communication. In other words, it has potential value. Specifically, we analyze popular hallucination benchmarks and reveal that hallucinated outputs display increased levels of narrativity and semantic coherence relative to veridical outputs. This finding reveals a tension in our usually dismissive understandings of confabulation. It suggests, counter-intuitively, that the tendency for LLMs to confabulate may be intimately associated with a positive capacity for coherent narrative-text generation.

Geologists have raised concerns about potential Chinese censorship and bias in a chatbot being developed with the backing of the International Union of Geological Sciences (IUGS), one of the world's largest scientific organisations and a Unesco partner. The GeoGPT chatbot is aimed at geoscientists and researchers, particularly in the global south, to help them develop their understanding of earth sciences by drawing on swaths of data and research on billions of years of the planet's history. It is an initiative from Deep-time Digital Earth (DDE), a largely Chinese-funded programme founded in 2019 to enhance international scientific cooperation and help countries to realise the UN's sustainable development goals. Part of the underlying AI for GeoGPT is Qwen, a large language model built by the Chinese tech company Alibaba. Responding to the article, DDE representatives Michael Stephenson, Hans Thybo, Chengshan Wang and Ishwaran Natarajan said the chatbot also used Meta's Llama, another large language model, and that during testing they had not noticed any state censorship, which they said was "unlikely" given that the system was "based entirely in geoscience information".

In a post-ChatGPT world, this paper explores the potential of leveraging scalable artificial intelligence for scientific discovery. We propose that scaling up artificial intelligence on high-performance computing platforms is essential to address such complex problems. This perspective focuses on scientific use cases like cognitive simulations, large language models for scientific inquiry, medical image analysis, and physics-informed approaches. The study outlines the methodologies needed to address such challenges at scale on supercomputers or the cloud and provides exemplars of such approaches applied to solve a variety of scientific problems. In light of ChatGPT's growing popularity, the transformative potential of AI in science becomes increasingly evident. Although a number of recent articles highlight the transformative power of AI in science [1, 2, 3], few provide specifics how to implement such methods at scale on supercomputers. Using ChatGPT as an archetype, we argue that the success of such complex AI models results from two primary advancements: (1) the development of the transformer architecture, (2) the ability to train on vast amounts of internet-scale data. This process represents a broader trend within the field of AI where combining massive amounts of training data with large-scale computational resources becomes the foundation of scientific breakthroughs. Several examples underscore the integral role of using large-scale computational resources and colossal amounts of data to achieve scientific breakthroughs. For instance, Khan et al. [4] used AI and large-scale computing for advanced models of black hole mergers, leveraging a dataset of 14 million waveforms on the Summit supercomputer. Riley et al. [5] made significant progress towards the understanding the physics of stratified fluid turbulence by being able to model the Prandtl number of seven, which represents ocean water at 20 Such simulations required being simulated using four trillion grid points, which required petabytes of storage [6].

Alignment of the language model with human preferences is a common approach to making a language model useful to end users. However, most alignment work is done in English, and human preference datasets are dominated by English, reflecting only the preferences of English-speaking annotators. Nevertheless, it is common practice to use the English preference data, either directly or by translating it into the target language, when aligning a multilingual language model. The question is whether such an alignment strategy marginalizes the preference of non-English speaking users. To this end, we investigate the effect of aligning Japanese language models with (mostly) English resources. In particular, we focus on evaluating whether the commonsense morality of the resulting fine-tuned models is aligned with Japanese culture using the JCommonsenseMorality (JCM) and ETHICS datasets. The experimental results show that the fine-tuned model outperforms the SFT model. However, it does not demonstrate the same level of improvement as a model fine-tuned using the JCM, suggesting that while some aspects of commonsense morality are transferable, others may not be.

Causal reasoning is a cornerstone of how humans interpret the world. To model and reason about causality, causal graphs offer a concise yet effective solution. Given the impressive advancements in language models, a crucial question arises: can they really understand causal graphs? To this end, we pioneer an investigation into language models' understanding of causal graphs. Specifically, we develop a framework to define causal graph understanding, by assessing language models' behaviors through four practical criteria derived from diverse disciplines (e.g., philosophy and psychology). We then develop CLEAR, a novel benchmark that defines three complexity levels and encompasses 20 causal graph-based tasks across these levels. Finally, based on our framework and benchmark, we conduct extensive experiments on six leading language models and summarize five empirical findings. Our results indicate that while language models demonstrate a preliminary understanding of causal graphs, significant potential for improvement remains. Our project website is at https://github.com/OpenCausaLab/CLEAR.

To what extent do LLMs align with human perceptions of safety? We study this question via *annotation alignment*, the extent to which LLMs and humans agree when annotating the safety of user-chatbot conversations. We leverage the recent DICES dataset (Aroyo et al., 2023), in which 350 conversations are each rated for safety by 112 annotators spanning 10 race-gender groups. GPT-4 achieves a Pearson correlation of $r = 0.59$ with the average annotator rating, higher than the median annotator's correlation with the average ($r=0.51$). We show that larger datasets are needed to resolve whether GPT-4 exhibits disparities in how well it correlates with demographic groups. Also, there is substantial idiosyncratic variation in correlation *within* groups, suggesting that race & gender do not fully capture differences in alignment. Finally, we find that GPT-4 cannot predict when one demographic group finds a conversation more unsafe than another.