DeBERTa-v3-base
PulseAugur coverage of DeBERTa-v3-base — every cluster mentioning DeBERTa-v3-base across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
New 184M-parameter safety classifier Semalith v1.4 outperforms Llama-Guard-3-8B on prompt injection
Researchers have introduced Semalith v1.4, a new safety classifier designed for large language models. This 184M-parameter model, built on DeBERTa-v3-base, excels at detecting prompt injection attacks and ensuring regul…
-
RAKSHAK system achieves 7th place in toxic intent classification at ACL 2026
Researchers have developed two systems, RAKSHAK and M1, for the GameTox Shared Task at ACL 2026, focusing on classifying toxic intent in World of Tanks chat. RAKSHAK, the primary system, utilizes a DeBERTa-v3-base model…
-
New pipeline targets gaming toxicity with neural-symbolic approach
Researchers have developed a novel neural-symbolic pipeline for detecting toxicity in online gaming chats, achieving a Macro F1 score of 0.6441 and an accuracy of 0.9062. Their system, which ranked third in Macro F1 and…
-
AI models improve human value detection using Schwartz theory geometry
Researchers have developed a new method called Schwartz-Geometry Decoding to improve human value detection in AI models. This approach leverages the theoretical structure of Schwartz values, which describes them as a co…
-
Study: Context, Moral Knowledge Aid Political Value Detection
A new study published on arXiv investigates the effectiveness of context, model size, and moral knowledge for detecting Schwartz values in political texts. Researchers found that while increased context improved supervi…
-
New ADE framework scales multi-anchor word representations for LLMs
Researchers have developed Adaptive Dictionary Embeddings (ADE), a new framework designed to scale multi-anchor word representations for large language models. ADE introduces techniques like Vocabulary Projection and Gr…