AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Decoupled Context Processing for Context Augmented Language Modeling

Zonglin Li, Ruiqi Guo, Sanjiv Kumar

NeurIPS 2022 · Large language models › Retrieval and knowledge · approach: Retrieval-augmented reasoning · 5 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add retrieved context to the language model (ablation)Decoupled Encoder-Decoder (base, large, XL) · C4 validationbits-per-byteImproved
use decoupled encoder-decoder cross attention instead of Retro chunked cross attentionRetro · C4 validationbits-per-byteImproved
use trained dense retriever instead of BM25 (ablation)Decoupled Encoder-Decoder · C4 validationbits-per-byteNo clear effect
train from scratch instead of fine-tuning pre-trained models (ablation)mT5 encoder-decoder · C4 validationbits-per-byteNo clear effect
add 100,000 extra steps with lower learning rate 2e-4 (ablation)Decoupled Encoder-Decoder · C4 validationbits-per-byteImproved
warm start with 100,000 steps on top 10% context-utility data (ablation)Large and XL models · C4 trainingtraining accuracyImproved
grounded context transfer versus non-grounded outputsDecoupled Encoder-Decoder · Natural Questionexact match accuracy: 51.5 -> 35.4Improved
remove passages containing the original output and rerun inference (ablation)Decoupled Encoder-Decoder · Natural Questionfraction of outputs changed but still grounded: 2,765 out of 3,354Mixed
precompute and cache context encodings instead of encoding at inferenceDecoupled Encoder-Decoder XL · Natural Question latencyretrieval time (ms): 200ms -> 66msImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy