CLINC150
PulseAugur coverage of CLINC150 — every cluster mentioning CLINC150 across labs, papers, and developer communities, ranked by signal.
6 day(s) with sentiment data
-
New framework evaluates machine unlearning for probabilistic decisions
Researchers have introduced DecisionUnlearn-Bench, a new empirical framework designed to evaluate machine unlearning for probabilistic decision models. This framework assesses how well models can forget specific data wh…
-
DeepOpen framework achieves high accuracy on banking and intent classification tasks
Researchers have introduced DeepOpen, a framework for decision-oriented intent classification tailored for specific datasets like BANKING77 and CLINC150. The system achieved a high accuracy of 94.12% on BANKING77 using …
-
ScaleDown CDMs tested: Specialized models show promise in classification, face scrutiny
ScaleDown has launched Constrained Decoding Models (CDMs), which are specialized models designed for specific tasks like classification, extraction, and summarization, aiming to be faster and cheaper than general-purpos…
-
Vektor's Desk AI agent boosts message routing accuracy to 90%
Vektor has released an update to its AI agent, Desk, version 2.0.1, significantly improving its automatic routing capabilities. The new version can now correctly route about 90% of messages to the appropriate mode, a su…
-
Jev decision model shows significant option-set dependence in tests
A new study published on arXiv investigates the option-set dependence of hosted decision models, specifically testing the Jev model. The research found that adding irrelevant alternatives to classification tasks signifi…
-
Cloudflare releases Clef-omni multimodal decision model
Cloudflare has released Clef-omni, a 30B multimodal mixture-of-experts model, on Hugging Face under an Apache 2.0 license. Unlike traditional LLMs, Clef-omni is designed for decision-making tasks, accepting text, images…
-
In-house agent router uses local models for efficient prompt routing
A company has developed an in-house agent router system designed to efficiently direct user prompts to the appropriate AI agent. The system utilizes two local models, bge-small and Kev-4B, to pre-screen prompts before t…
-
Cloudflare launches open-source decision models Clef and Clef-flash
Cloudflare has released two open-source decision models, Clef and Clef-flash, designed for tasks like classification and choice selection without generating text. Built on Qwen3.8-27B and Qwen3.5-9B respectively, these …
-
New Jev Model Property Enables Zero-Label Prediction and Compute Allocation
Researchers have introduced a new property in prefill-only decision models, inspired by the Jev model, which allows for zero-label prediction and inference-time compute allocation. This property enables models to predic…
-
Convai Innovations' Laya decision engine detailed in developer guide
Laya, an open-source decision engine from Convai Innovations, is detailed in a developer's guide. This non-autoregressive System 1 model uses a 421-million-parameter encoder to process text and questions in a single pas…
-
New TPKD method enhances AI model knowledge transfer beyond imitation
Researchers have developed a new method called task-preserving knowledge distillation (TPKD) that aims to improve the efficiency of knowledge transfer between AI models. Unlike traditional methods that focus on exact im…
-
New Lexical Confidence Method Enhances AI Intent Routing Reliability
Researchers have developed a new method called Signed Lexical Confidence (SLC) to improve the reliability of AI intent routing. This technique combines a classifier's confidence score with evidence from a lexical model,…
-
LLMs vs. Fine-Tuned NLU: New Framework Guides Intent Detection Choices
A new research paper explores when large language models (LLMs) are a suitable replacement for fine-tuned Natural Language Understanding (NLU) models in conversational systems. The study found that while fine-tuned mode…
-
New ERGO method optimizes text classification by learning from errors
Researchers have developed a new method called Error-Guided Optimization (ERGO) for text classification tasks. ERGO iteratively diagnoses classification failures on batches of data and generates targeted decision rules …
-
New method improves out-of-scope intent detection using MiniLM embeddings
Researchers have developed a novel multi-cluster boundary learning method for out-of-scope (OOS) intent detection, utilizing MiniLM embeddings. This approach addresses challenges in traditional OOS detection, such as de…
-
New OCRR benchmark measures AI model recovery from distribution shift via corrections
Researchers have introduced OCRR, a new benchmark designed to evaluate how well machine learning systems can recover from distribution shifts using online corrections. Unlike static benchmarks, OCRR simulates real-world…