As the usage of large language models (LLMs) grows, performing efficient inference with these models becomes increasingly important. While speculative decoding has recently emerged as a promising direction for speeding up inference, existing methods are limited in their ability to scale to larger speculation budgets, and adapt to different hyperparameters and hardware. This paper introduces Sequoia, a scalable, robust, and hardware-aware algorithm for speculative decoding. To attain better scalability, Sequoia introduces a dynamic programming algorithm to find the optimal tree structure for the speculated tokens. To achieve robust speculative performance, Sequoia uses a novel sampling and verification method that outperforms prior work across different decoding temperatures. Finally, Sequoia introduces a hardware-aware tree optimizer that maximizes speculative performance by automatically selecting the token tree size and depth for a given hardware platform. Evaluation shows that Sequoia improves the decoding speed of Llama2-7B, Llama2-13B, and Vicuna-33B on an A100 by up to $4.04\times$, $3.73\times$, and $2.27\times$. For offloading setting on L40, Sequoia achieves as low as 0.56 s/token for exact Llama2-70B inference latency, which is $9.96\times$ on our optimized offloading system (5.6 s/token), $9.7\times$ than DeepSpeed-Zero-Inference, $19.5\times$ than Huggingface Accelerate.

Knowledge editing aims to inject knowledge updates into language models to keep them correct and up-to-date. However, its current evaluation strategies are notably impractical: they solely update with well-curated structured facts (triplets with subjects, relations, and objects), whereas real-world knowledge updates commonly emerge in unstructured texts like news articles. In this paper, we propose a new benchmark, Unstructured Knowledge Editing (UKE). It evaluates editing performance directly using unstructured texts as knowledge updates, termed unstructured facts. Hence UKE avoids the laborious construction of structured facts and enables efficient and responsive knowledge editing, becoming a more practical benchmark. We conduct extensive experiments on newly built datasets and demonstrate that UKE poses a significant challenge to state-of-the-art knowledge editing methods, resulting in their critical performance declines. We further show that this challenge persists even if we extract triplets as structured facts. Our analysis discloses key insights to motivate future research in UKE for more practical knowledge editing.

Alignment in artificial intelligence pursues the consistency between model responses and human preferences as well as values. In practice, the multifaceted nature of human preferences inadvertently introduces what is known as the "alignment tax" -a compromise where enhancements in alignment within one objective (e.g.,harmlessness) can diminish performance in others (e.g.,helpfulness). However, existing alignment techniques are mostly unidirectional, leading to suboptimal trade-offs and poor flexibility over various objectives. To navigate this challenge, we argue the prominence of grounding LLMs with evident preferences. We introduce controllable preference optimization (CPO), which explicitly specifies preference scores for different objectives, thereby guiding the model to generate responses that meet the requirements. Our experimental analysis reveals that the aligned models can provide responses that match various preferences among the "3H" (helpfulness, honesty, harmlessness) desiderata. Furthermore, by introducing diverse data and alignment goals, we surpass baseline methods in aligning with single objectives, hence mitigating the impact of the alignment tax and achieving Pareto improvements in multi-objective alignment.

Social media platforms are daily exhibiting millions of events. To preliminarily predict the mainstream public reaction to these events, we study trendy response prediction to automatically generate top-liked user replies to social media events. While previous works focus on generating responses without factoring in popularity, we propose Popularity-Aligned Language Models (PopALM) to distinguish responses liked by a larger audience through reinforcement learning. Recognizing the noisy labels from user "likes", we tailor-make curriculum learning in proximal policy optimization (PPO) to help models capture the essential samples for easy-to-hard training. In experiments, we build a large-scale Weibo dataset for trendy response prediction, and its results show that PopALM can help boost the performance of advanced language models.

This paper investigates the gap in representation powers of Recurrent Neural Networks (RNNs) and Transformers in the context of solving algorithmic problems. We focus on understanding whether RNNs, known for their memory efficiency in handling long sequences, can match the performance of Transformers, particularly when enhanced with Chain-of-Thought (CoT) prompting. Our theoretical analysis reveals that CoT improves RNNs but is insufficient to close the gap with Transformers. A key bottleneck lies in the inability of RNNs to perfectly retrieve information from the context, even with CoT: for several tasks that explicitly or implicitly require this capability, such as associative recall and determining if a graph is a tree, we prove that RNNs are not expressive enough to solve the tasks while Transformers can solve them with ease. Conversely, we prove that adopting techniques to enhance the in-context retrieval capability of RNNs, including Retrieval-Augmented Generation (RAG) and adding a single Transformer layer, can elevate RNNs to be capable of solving all polynomial-time solvable problems with CoT, hence closing the representation gap with Transformers.

Three progressive US outlets – the Intercept, Raw Story and AlterNet – filed suits in Manhattan federal court on Wednesday, demanding compensation from the tech companies. "It's important to democracy that a diverse array of news sites continue to thrive. OpenAI's violations, if not checked, will further decimate the news industry, and with it, the critical news reporters who affect positive change." The Intercept's suit lists both OpenAI and its most prominent investor Microsoft as defendants, while the joint suit filed by Raw Story and AlterNet only lists OpenAI. The complaints are otherwise nearly identical, and the law firm Loevy & Loevy is representing all three outlets in the suits.

More than half the world's population, including India, the US and UK, will have the chance to vote for new governments in 2024 THE biggest election year in the history of the world is under way, and we have just had our first glimpse at how artificial intelligence is being used by the shadowy world of government-backed hackers. These groups could have significant impacts on democratic processes through hacking, disinformation or leaking sensitive information.

February was shaping up to be a banner month for Google's ambitious artificial intelligence strategy. The company rebranded its chatbot as Gemini and released two major product upgrades to better compete with rivals on all sides in the high-stakes AI arms race. In the midst of all that, Google also began allowing Gemini users to generate realistic-looking images of people. Not many noticed the feature at first. Other companies like OpenAI already offer tools that let users quickly make images of people that can then be used for marketing, art and brainstorming creative ideas.

This is the future of AI video. When videos like these are made completely by artificial intelligence. None of these videos depict real people, places or events. At first glance, the images amaze and confound: A woman strides along a city street alive with pedestrians and neon lights. A car kicks up a cloud of dust on a mountain road.

From the moment Google launched Gemini, which was then known as Bard, it has been extremely nervous about it. Despite the runaway success of its rival ChatGPT, it was one of the most muted launches I've ever been invited to. Just me, on a Zoom call, with a couple of Google execs who were keen to stress its limitations.