Despite growing interest in AI agents across industry and academia, their execution in an environment is often slow, hampering training, evaluation, and deployment. For example, a game of chess between two state-of-the-art agents may take hours. A critical bottleneck is that agent behavior unfolds sequentially: each action requires an API call, and these calls can be time-consuming. Inspired by speculative execution in microprocessors and speculative decoding in LLM inference, we propose speculative actions, a lossless framework for general agentic systems that predicts likely actions using faster models, enabling multiple steps to be executed in parallel. We evaluate this framework across three agentic environments: gaming, e-commerce, web search, and a "lossy" extension for an operating systems environment. In all cases, speculative actions achieve substantial accuracy in next-action prediction (up to 55%), translating into significant reductions in end-to-end latency. Moreover, performance can be further improved through stronger guessing models, top-K action prediction, multi-step speculation, and uncertainty-aware optimization, opening a promising path toward deploying low-latency agentic systems in the real world.

Analyzing large datasets requires responsive query execution, but executing SQL queries on massive datasets can be slow. This paper explores whether query execution can begin even before the user has finished typing, allowing results to appear almost instantly. We propose SpeQL, a system that leverages Large Language Models (LLMs) to predict likely queries based on the database schema, the user's past queries, and their incomplete query. Since exact query prediction is infeasible, SpeQL speculates on partial queries in two ways: 1) it predicts the query structure to compile and plan queries in advance, and 2) it precomputes smaller temporary tables that are much smaller than the original database, but are still predicted to contain all information necessary to answer the user's final query. Additionally, SpeQL continuously displays results for speculated queries and subqueries in real time, aiding exploratory analysis. A utility/user study showed that SpeQL improved task completion time, and participants reported that its speculative display of results helped them discover patterns in the data more quickly. In the study, SpeQL improves user's query latency by up to $289\times$ and kept the overhead reasonable, at $\$4$ per hour.

Improving time-to-first-token (TTFT) is an essentially important objective in modern large language model (LLM) inference engines. Because optimizing TTFT directly results in higher maximal QPS and meets the requirements of many critical applications. However, boosting TTFT is notoriously challenging since it is purely compute-bounded and the performance bottleneck shifts from the self-attention to the MLP part. We present SpecPrefill, a training free framework that accelerates the inference TTFT for both long and medium context queries based on the following insight: LLMs are generalized enough to still preserve the quality given only a carefully chosen subset of prompt tokens. At its core, SpecPrefill leverages a lightweight model to speculate locally important tokens based on the context. These tokens, along with the necessary positional information, are then sent to the main model for processing. We evaluate SpecPrefill with a diverse set of tasks, followed by a comprehensive benchmarking of performance improvement both in a real end-to-end setting and ablation studies. SpecPrefill manages to serve Llama-3.1-405B-Instruct-FP8 with up to $7\times$ maximal end-to-end QPS on real downstream tasks and $7.66\times$ TTFT improvement during benchmarking.

Large Language Models (LLMs) frequently suffer from inefficiencies, largely attributable to the discord between the requirements of auto-regressive decoding and the architecture of contemporary GPUs. Recently, regressive lightweight speculative decoding has garnered attention for its notable efficiency improvements in text generation tasks. This approach utilizes a lightweight regressive draft model, like a Recurrent Neural Network (RNN) or a single transformer decoder layer, leveraging sequential information to iteratively predict potential tokens. Specifically, RNN draft models are computationally economical but tend to deliver lower accuracy, while attention decoder layer models exhibit the opposite traits. This paper presents Clover-2, an advanced iteration of Clover, an RNN-based draft model designed to achieve comparable accuracy to that of attention decoder layer models while maintaining minimal computational overhead. Clover-2 enhances the model architecture and incorporates knowledge distillation to increase Clover's accuracy and improve overall efficiency. We conducted experiments using the open-source Vicuna 7B and LLaMA3-Instruct 8B models. The results demonstrate that Clover-2 surpasses existing methods across various model architectures, showcasing its efficacy and robustness.

One approach to squaring this circle is speculative decoding, where a smaller draft model or speculator is trained This technical report describes the design and training to predict multiple tokens given a sequence of input. These of novel speculative decoding draft models, for accelerating speculative tokens are produced with low cost, and lower the inference speeds of large language models in a accuracy than the base LLM. However, we can leverage production environment. By conditioning draft predictions GPU parallelism during the LLM forward pass to evaluate on both context vectors and sampled tokens, we can train the output for each of these new tokens with minimal additional our speculators to efficiently predict high-quality n-grams, overhead. Then, by comparing the outputs to the which the base model then accepts or rejects. This allows us speculated inputs, we can accept all the predicted tokens to effectively predict multiple tokens per inference forward that match the output of the base model, while rejecting all pass, accelerating wall-clock inference speeds of highly optimized those that don't. In this way we can predict multiple tokens base model implementations by a factor of 2-3x. We per LLM forward pass at minimal extra cost. A deeper explore these initial results and describe next steps for further explanation of speculative decoding can be found in [3, 6].

Large language models (LLMs) suffer from low efficiency as the mismatch between the requirement of auto-regressive decoding and the design of most contemporary GPUs. Specifically, billions to trillions of parameters must be loaded to the GPU cache through its limited memory bandwidth for computation, but only a small batch of tokens is actually computed. Consequently, the GPU spends most of its time on memory transfer instead of computation. Recently, parallel decoding, a type of speculative decoding algorithms, is becoming more popular and has demonstrated impressive efficiency improvement in generation. It introduces extra decoding heads to large models, enabling them to predict multiple subsequent tokens simultaneously and verify these candidate continuations in a single decoding step. However, this approach deviates from the training objective of next token prediction used during pre-training, resulting in a low hit rate for candidate tokens. In this paper, we propose a new speculative decoding algorithm, Clover, which integrates sequential knowledge into the parallel decoding process. This enhancement improves the hit rate of speculators and thus boosts the overall efficiency. Clover transmits the sequential knowledge from pre-speculated tokens via the Regressive Connection, then employs an Attention Decoder to integrate these speculated tokens. Additionally, Clover incorporates an Augmenting Block that modifies the hidden states to better align with the purpose of speculative generation rather than next token prediction. The experiment results demonstrate that Clover outperforms the baseline by up to 91% on Baichuan-Small and 146% on Baichuan-Large, respectively, and exceeds the performance of the previously top-performing method, Medusa, by up to 37% on Baichuan-Small and 57% on Baichuan-Large, respectively.

Non-Fungible Tokens (NFTs) promise to revolutionize how content creators (e.g., artists) price and sell their work. One core feature of NFTs is the option to embed creator royalties which earmark a percentage of future sale proceeds to creators, each time their NFTs change hands. As popular as this feature is in practice, its utility is often questioned because buyers, the argument goes, simply ``price it in at the time of purchase''. As intuitive as this argument sounds, it is incomplete. We find royalties can add value to creators in at least three distinct ways. (i) Risk sharing: when creators and buyers are risk sensitive, royalties can improve trade by splitting the risks associated with future price volatility; (ii) Dynamic pricing: in the presence of information asymmetry, royalties can extract more revenues from better-informed speculators over time, mimicking the benefits of ``dynamic pricing''; (iii) Price discrimination: when creators sell multi-unit NFT collections, royalties can better capture value from heterogeneous buyers. Our results suggest creator royalties play an important and sometimes overlooked role in the economics of NFTs.

Deep learning technology has rattled the global financial industry in both positive and negative ways. On the one hand, deep learning technology has considerably improved market efficiency. Tomiwa, a big data author and expert, claims to have beaten the stock market average over the past ten years with a program that he developed with Python. The same kind of program could be used by Forex or derivative traders. One of the biggest downsides, though, is that it has giving larger institutional traders with deep pockets an even stronger advantage.

In this paper, an event-centric commodity trading simulation powered by the multiagent framework is presented. The purpose of this simulation platform is for training novice traders. The simulation is progressed by announcing news events that affect various aspects of the commodity supply chain. Upon receiving these events, market agents that play the roles of producers, consumers, and speculators would adjust their views on the market and act accordingly. Their actions would be based on their roles and also their private information, and collectively they shape the market dynamics. This simulation has been effectively deployed for several training sessions. We will present the underlying technologies that are employed and discuss the practical significance of such platform.