cd /news/large-language-models/learning-to-translate-from-soft-to-h… · home topics large-language-models article
[ARTICLE · art-16069] src=arxiv.org pub= topic=large-language-models verified=true sentiment=↑ positive

Learning to Translate from Soft to Hard LLM Prompts

Researchers developed a translation model that converts soft prompts—parameter-efficient but uninterpretable LLM adaptations—into natural language text. The translated prompts outperformed existing interpretability methods like InSPEcT on multiple datasets, producing more fluent and accurate verbalizations. This approach enables soft prompts optimized on small open-source models to be deployed as portable text prompts on larger closed-API models, where they can exceed the original soft prompt's performance and even rival few-shot learning.

read1 min publishedMay 28, 2026

arXiv:2605.27642v1 Announce Type: new Abstract: Soft prompt tuning is a parameter-efficient method for adapting LLMs to specific tasks, but suffers from a lack of interpretability. Building on recent work on interpreting soft prompts (Ramati et al., 2024), we explore how training a dedicated soft prompt to natural language translation model can yield higher translation quality. In particular, in both quantitative and qualitative comparisons on multiple Datasets of Datasets (DoDs), we demonstrate that our translator produces fluent, accurate verbalizations that outperforms existing training-free methods like InSPEcT. In addition to advancing interpretability, our work suggests a promising downstream application: soft prompts optimized on small, open-source models can be translated into portable text prompts that, when deployed on larger closed-API models, exceed the performance of the original soft prompt and, in some cases, even few-shot learning.

── more in #large-language-models 4 stories · sorted by recency
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/learning-to-translat…] indexed:0 read:1min 2026-05-28 ·