Brieftide
Topic hub
Retrieval-Augmented Models
Covers techniques that combine retrieval and models, including long-term conversational memory, personalization, robustness, and modular model composition.
100 briefs
Retrieval-Augmented Models · Page 4
- Anthropic and Stanford: why larger LLMs learn rare skillsThe Brieftide
- LLM Research Papers 2026 (Jan–May): Curated list and trendsThe Brieftide
- ChatGPT Dreaming V3: OpenAI memory rollout in US, wider launchThe Brieftide
- Meta AI support bot allowed Instagram account takeoversThe Brieftide
- Mellum2 by JetBrains: 12B MoE model with 2.5B active paramsThe Brieftide
- Reachy Mini goes fully local: run speech-to-speech stackThe Brieftide
- Google Search: Gemini 3.5 Flash, AI agents and smart Search boxThe Brieftide
- Ettin reranker family: six CrossEncoders, MTEB retrievalThe Brieftide
- Gemini Omni Flash launch: Google DeepMind's multimodal video AIThe Brieftide
- Gemma 4 and new LLM designs: KV sharing, PLE, compressed attentionThe Brieftide
- Granite Embedding Multilingual R2: 97M tops sub-100M MTEB scoresThe Brieftide
- NVIDIA Nemotron 3 Nano Omni launch: long-context multimodal AIThe Brieftide
- DeepSeek-V4: million-token context release from Hugging FaceThe Brieftide
- GRASP: Gradient Planning for World Models at Long HorizonsThe Brieftide
- Gemini Robotics ER 1.6 release: improved embodied reasoningThe Brieftide
- Sentence Transformers launches multimodal embeddings and rerankersThe Brieftide
- Attention variants in LLMs: MHA, GQA, MLA, sparse & hybridThe Brieftide
- Gemini 3.1 Pro release: DeepMind's model for complex tasksThe Brieftide
- DeepSeek V3: 671B model, MLA and MoE architectural choices, 2025The Brieftide
- KV Cache in LLMs: How Sebastian Raschka Implements ItThe Brieftide
- SecAlign and StruQ: Berkeley AI defenses cut prompt-injectionThe Brieftide