AI Briefing#language-model
The latest AI and developer news about #language-model, with the original source and a short summary.
The latest AI and developer news about #language-model, with the original source and a short summary.
The model uses caller-defined tools to decide whether to send a message or pass, supporting group chat dynamics.
Sparse data is optimally reused 15–20 times, leading to the proposal of a scaling law that reflects this.
Mimir, a new 1.7B-scale model supporting English and Danish, has been released, demonstrating performance that surpasses Qwen and Gemma.
DiffusionGemma accelerates text generation by iteratively refining 256 tokens in parallel.
A next-generation language model, celeris-1, has been unveiled, applying Diffusion technology instead of the traditional Autoregressive approach to revolutionize speed.
Anthropic unveiled an LLM reasoning tool called Schema, which achieved 99% performance on the ARC-AGI benchmark.
Codex's GPT-5.6 offers three models—Sol, Terra, and Luna—that you can choose based on task complexity.
The RTCA workshop at NeurIPS 2026, covering research on real-time multimodal conversational agents, has opened its call for papers.
A review of research trends in language models combining knowledge information, centered on ACL 2022.
Using Parameter Decomposition technology, they succeeded in precisely removing only the model's German language ability while preserving its abilities in other languages.
The Estonian Language Institute (EKI) has released a new benchmark that evaluates LLM performance in a non-mainstream language and resistance to propaganda.
NVIDIA has released a Diffusion language model family that delivers inference speeds up to 6x faster than existing AR models.
EMO has been released, maintaining performance even with only some experts used, thanks to document-level routing.
IBM has unveiled the Granite 4.1 model series, optimized for multilingual generation, coding, RAG, and AI assistant workflows.
Talkie, a 13B model trained exclusively on pre-1931 text, has been introduced.
Flipping just a few sign bits in the weights can effectively collapse vision and language models.
To overcome the limitations of existing attribute-based similarity calculations, Oliveyoung developed a semantic product similarity model that leverages user search logs.
This article explains the principles of MLM and CLM, which are core to BERT and GPT2, and introduces how to build domain-specific models by directly implementing a Transformer encoder in PyTorch.
Harambee boosted sign-up conversion by 20% with an AI21-based chatbot.
This provides a technical guide on improving Wav2Vec2's speech recognition performance using an n-gram language model.
AI21 Labs has launched **Wordtune Read**, a document summarization tool.
AI21 Labs has unveiled its developer platform AI21 Studio and Jurassic-1.
The DeDLOC algorithm has been released, which trains language models by leveraging the resources of multiple users even in slow internet network environments.