SemEval-2026
PulseAugur coverage of SemEval-2026 — every cluster mentioning SemEval-2026 across labs, papers, and developer communities, ranked by signal.
-
Researchers develop lightweight method to detect LLM-generated code
Researchers have developed a lightweight method for detecting code generated by large language models (LLMs). Their approach, presented for SemEval-2026 Task 13, utilizes stylometric signals and ratio-based features tha…
-
RaguTeam wins SemEval-2026 LLM task with judge-orchestrated ensemble
RaguTeam has developed a winning system for the SemEval-2026 Task 8, which focuses on faithful multi-turn response generation. Their approach utilizes a heterogeneous ensemble of seven large language models, with a GPT-…
-
Neuro-symbolic AI advances offer explainability and reasoning beyond pure neural networks
Researchers are developing neuro-symbolic AI models that combine neural networks with symbolic reasoning to improve explainability and performance. Gyan, a novel non-transformer architecture, aims to overcome limitation…
-
SemEval-2026 task evaluates LLM knowledge across 30+ low-resource languages
A new shared task, SemEval-2026 Task 7, has been introduced to evaluate the adaptability of language models and NLP systems across diverse languages and cultures. The task utilizes an extended version of the BLEnD bench…
-
Archaeology team fine-tunes code models for AI-generated code detection
Researchers from the team Archaeology have developed a system for detecting AI-generated code, participating in SemEval-2026 Task 13. Their approach involves fine-tuning several pre-trained code models, including CodeBE…
-
CLaC system uses LLMs and encoders for political discourse clarity detection
Researchers presented a system for SemEval-2026 Task 6, focusing on detecting clarity and evasion in political discourse. Their approach involved comparing fine-tuned encoders with prompt-based large language models (LL…
-
Teams leverage LLMs and ensemble methods for multilingual online polarization detection at SemEval-2026
Researchers have developed systems for SemEval-2026 Task 9, a multilingual polarization detection challenge across 22 languages. One approach fine-tuned Gemma 3 models using Low-Rank Adaptation (LoRA) and augmented data…
-
LLM finetuning system achieves 85th percentile in conspiracy detection task
Researchers developed an mdok-style system for SemEval-2026 Task 10, which focuses on detecting conspiracy beliefs in Reddit comments. The system employed data augmentation and self-training techniques to fine-tune the …
-
H-RAG paper details hierarchical retrieval for multi-turn RAG conversations
Researchers have introduced H-RAG, a novel hierarchical retrieval-augmented generation system designed for multi-turn conversational AI. This approach separates retrieval into fine-grained child chunks and parent-level …
-
SG-UniBuc-NLP uses RoBERTa with chunking for political evasion detection
Researchers from SG-UniBuc-NLP developed a system for SemEval-2026 Task 6, which aims to detect political question evasions in English interviews. Their approach utilizes a Multi-Head RoBERTa model combined with a chunk…
-
SemEval-2026 task focuses on narrative story similarity and representation learning
Researchers have introduced a new task, NSNRL, for evaluating narrative story similarity and representation learning. The task frames similarity as a binary classification problem, determining which of two stories is mo…
-
UKP_Psycontrol wins SemEval-2026 Task 2 for modeling text-based emotion dynamics
Researchers from UKP_Psycontrol have developed a system for SemEval-2026 Task 2, which focuses on predicting affective states and their changes from user-generated text. Their approach combined large language model prom…