TALM: Tool Augmented Language Models

Parisi, Aaron; Zhao, Yao; Fiedel, Noah

doi:10.48550/arxiv.2205.12255

Cited by 10 publications

(15 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We first evaluate how augmenting the algorithmic prompt into the chain-of-thought would affect the performance. We then show how algorithmic prompt can be used as tool use (Parisi et al, 2022), where a model queries another source for a particular type of information.…”

Section: Using Skills As Toolsmentioning

confidence: 99%

See 1 more Smart Citation

Teaching Algorithmic Reasoning via In-context Learning

Zhou¹,

Nova²,

Larochelle³

et al. 2022

Preprint

View full text Add to dashboard Cite

Large language models (LLMs) have shown increasing in-context learning capabilities through scaling up model and data size. Despite this progress, LLMs are still unable to solve algorithmic reasoning problems. While providing a rationale with the final answer has led to further improvements in multi-step reasoning problems, Anil et al. (2022) showed that even simple algorithmic reasoning tasks such as parity are far from solved. In this work, we identify and study four key stages for successfully teaching algorithmic reasoning to LLMs: (1) formulating algorithms as skills, (2) teaching multiple skills simultaneously (skill accumulation), (3) teaching how to combine skills (skill composition) and ( 4) teaching how to use skills as tools. We show that it is possible to teach algorithmic reasoning to LLMs via in-context learning, which we refer to as algorithmic prompting. We evaluate our approach on a variety of arithmetic and quantitative reasoning tasks, and demonstrate significant boosts in performance over existing prompting techniques. In particular, for long parity, addition, multiplication and subtraction, we achieve an error reduction of approximately 10x, 9x, 5x and 2x respectively compared to the best available baselines.

show abstract

Section: Using Skills As Toolsmentioning

confidence: 99%

“…Creswell and Shanahan (2022) uses a similar multi-model and multi-prompt strategy in order to separate out selection from inference in reasoning problems. This approach can be considered a form of tool use (Parisi et al, 2022), where a model queries another source for a particular type of information.…”

Section: Algorithmic Prompt As Tool Usementioning

confidence: 99%

Teaching Algorithmic Reasoning via In-context Learning

Zhou¹,

Nova²,

Larochelle³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Recently, many works extend the capabilities of LLMs by giving them access to external models, tools and APIs [25], [26], [27], [28]. Tool usage can also be combined with reasoning techniques such as CoT prompting [9] to significantly improve planning [29].…”

Section: B Orchestrating Robot Behavior With Llmsmentioning

confidence: 99%

Deep Episodic Memory for Verbalization of Robot Experience

Brmann

Peller-Konrad

Constantin

et al. 2021

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

Natural-language dialog is key for intuitive humanrobot interaction. It can be used not only to express humans' intents, but also to communicate instructions for improvement if a robot does not understand a command correctly. Of great importance is to endow robots with the ability to learn from such interaction experience in an incremental way to allow them to improve their behaviors or avoid mistakes in the future. In this paper, we propose a system to achieve incremental learning of complex behavior from natural interaction, and demonstrate its implementation on a humanoid robot. Building on recent advances, we present a system that deploys Large Language Models (LLMs) for high-level orchestration of the robot's behavior, based on the idea of enabling the LLM to generate Python statements in an interactive console to invoke both robot perception and action. The interaction loop is closed by feeding back human instructions, environment observations, and execution results to the LLM, thus informing the generation of the next statement. Specifically, we introduce incremental prompt learning, which enables the system to interactively learn from its mistakes. For that purpose, the LLM can call another LLM responsible for code-level improvements of the current interaction based on human feedback. The improved interaction is then saved in the robot's memory, and thus retrieved on similar requests. We integrate the system in the robot cognitive architecture of the humanoid robot ARMAR-6 and evaluate our methods both quantitatively (in simulation) and qualitatively (in simulation and real-world) by demonstrating generalized incrementally-learned knowledge.

show abstract

“…Digital environments. This includes interacting with games Côté et al, 2019;Shridhar et al, 2020;Liu et al, 2023d), APIs Yao et al, 2022b;Parisi et al, 2022;, and websites (Shi et al, 2017;Nakano et al, 2021;Yao et al, 2022a;Gur et al, 2023;Deng et al, 2023) as well as general code execution Le et al, 2022;Ni et al, 2023). Such digital grounding is cheaper and faster than physical or human interaction.…”

Section: Grounding Actionsmentioning

confidence: 99%

“…It is thus a convenient testbed for language agents and has been studied with increasing intensity in recent years. In particular, for NLP tasks that require augmentation of external knowledge or computation, stateless digital APIs (e.g., search, calculator, translator) are often packaged as "tools" (Parisi et al, 2022;Qin et al, 2023), which can be viewed as special "single-use" digital environments.…”

Section: Grounding Actionsmentioning

confidence: 99%

Show or Tell? Exploring when (and why) teaching with language outperforms demonstration

Sumers¹,

Ho²,

Hawkins³

et al. 2021

Preprint

View full text Add to dashboard Cite

People use a wide range of communicative acts, from concrete demonstrations to abstract language. What are the strengths and weaknesses of such different modalities? We present a series of real-time, multi-player experiments asking participants to teach (Boolean) concepts using either demonstrations or language. Our first experiment (N = 454) manipulated the complexity of the concept, finding that linguistic (but not demonstrative) teaching enables high-fidelity transmission of more complex concepts. Why, then, do humans use both demonstrations and language? As a form of conventionalized communication, language relies on shared context between speaker and listener, whereas demonstrations are inherently grounded in the world. We hypothesized linguistic communication would be more sensitive to perturbations of shared context than demonstrations. Our second experiment (N = 568) manipulated teachers’ ability to see the features that defined the concept. This restriction severely impaired linguistic (but not demonstrative) teaching. Our comparative approach confirms language relies on shared context to permit high bandwidth communication; in contrast, demonstrations are lower-bandwidth but more robust.

show abstract

TALM: Tool Augmented Language Models

Cited by 10 publications

References 0 publications

Teaching Algorithmic Reasoning via In-context Learning

Teaching Algorithmic Reasoning via In-context Learning

Deep Episodic Memory for Verbalization of Robot Experience

Show or Tell? Exploring when (and why) teaching with language outperforms demonstration

Contact Info

Product

Resources

About