English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Large language models (LLMs) exhibit in-context learning abilities whichenable the same model to perform several tasks without any task-specifictraining. In contrast, traditional adaptation approaches, such as fine-tuning,modify the underlying models for each specific task. In-context learning,however, consistently underperforms task-specific tuning approaches even whenpresented with the same examples. While most existing approaches (e.g., promptengineering) focus on the LLM's learned representations to patch thisperformance gap, our analysis actually reveal that LLM representations containsufficient information to make good predictions. As such, we focus on the LLM'sreasoning abilities and demonstrate that this performance gap exists due totheir inability to perform simple probabilistic reasoning tasks. This raises anintriguing question: Are LLMs actually capable of learning how to reason in atask-agnostic manner? We answer this in the affirmative and propose TART whichgenerically improves an LLM's reasoning abilities using a synthetically trainedTransformer-based reasoning module. TART trains this reasoning module in atask-agnostic manner using only synthetic logistic regression tasks andcomposes it with an arbitrary real-world pre-trained model without anyadditional training. With a single inference module, TART improves performanceacross different model families (GPT-Neo, Pythia, BLOOM), model sizes (100M -6B), tasks (14 NLP binary classification tasks), and even across differentmodalities (audio and vision). Additionally, on the RAFT Benchmark, TARTimproves GPT-Neo (125M)'s performance such that it outperforms BLOOM (176B),and is within 4% of GPT-3 (175B). Our code and models are available athttps://github.com/HazyResearch/TART .

TART: A plug-and-play Transformer module for task-agnostic reasoning