AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Can Unconditional Language Models Recover Arbitrary Sentences?

Nishant Subramani, Samuel Bowman, Kyunghyun Cho

NeurIPS 2019 · Large language models › Language, safety and interpretability · approach: Sign and cross-lingual translation · 15 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

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
Increase reparametrized-space dimension from 512 to 4096 (ablation)Large LSTM trained on 50M Gigaword sentences · Gigaword held-out sentencesBLEU: 54.8 -> 99.8Improved
Increase reparametrized-space dimension from 512 to 2048 (ablation)Medium LSTM trained on 50M Gigaword sentences · Gigaword held-out sentencesBLEU: 41.1 -> 91.2Improved
Increase reparametrized-space dimension from 512 to 8192 (ablation)Small LSTM trained on 50M Gigaword sentences · Gigaword held-out sentencesBLEU: 40.0 -> 81.1Improved
Increase language-model training data from 10M to 50M sentences (ablation)Fixed LSTM language models with reparametrized sentence conditioning · Gigaword held-out sentencesrecoverabilityImproved
Use Adam instead of nonlinear conjugate gradient for forward estimation (ablation)Reparametrized sentence-space optimization · Gigaword held-out sentencesBLEU: do not exceed 1.0 BLEUHurt
Increase beam width from 5 to 10 (ablation)Large LSTM, |Z|=4096 · Gigaword held-out sentencesBLEU: 99.8 -> 99.8No clear effect
Increase beam width from 5 to 20 (ablation)Large LSTM, |Z|=4096 · Gigaword held-out sentencesBLEU: 99.8 -> 99.5No clear effect
Use |Z| greater than the model dimension (ablation)Well-trained medium and large LSTMs · Gigaword held-out sentencesrecoverabilityNo clear effect
Increase sentence lengthReparametrized sentence-space recovery · Gigaword held-out sentencesrecoverabilityHurt
Condition the medium and large models on optimized sentence vectors with |Z| at least the model dimensionFixed pretrained LSTM language models · IWSLT16 English-to-German validation sentencesBLEUImproved
Recover randomly sampled token sequences longer than 25 subword unitsLarge LSTM with reparametrized sentence conditioning · Random sequencesBLEUHurt

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