AI Briefing#llm
The latest AI and developer news about #llm, with the original source and a short summary.
The latest AI and developer news about #llm, with the original source and a short summary.
Anthropic has launched Claude services for consumers and businesses in Brazil.
LangChain has released its annual report covering 2024 AI development trends and LLM infrastructure usage.
FuriosaAI's AI inference chip **RNGD** demonstrated outstanding throughput and power efficiency on **Llama 3.1** models.
IBM and researchers from major universities have unveiled Bamba-9B, a hybrid Mamba2 model that maximizes inference efficiency.
OpenAI has unveiled new tools for developers, including API support for the reasoning model **o1** and improvements to the **Realtime API**.
TII has released the Falcon 3 open model family, featuring under-10B-parameter models with enhanced math, coding, and reasoning capabilities.
Synthetic Data Generator, a no-code tool that lets you create custom datasets using only natural language descriptions, has been released.
xAI has made an improved Grok-2 available for free to 𝕏 users and cut API pricing.
Zalando adopted **GPT-4o mini** to strengthen multilingual support and personalized recommendation features, boosting both customer experience and operational efficiency at the same time.
OpenAI has launched the ChatGPT Pro plan, offering high-performance models and o1 pro mode for $200 per month.
With an average score of 75.1 on Korean benchmarks, surpassing GPT-4o (61.9), the model adopts Native resolution and the C-Abstractor architecture.
This introduces CFM's case of using LLM-based data labeling to boost small model performance while cutting operational costs by up to 80x.
FuriosaAI is joining UEC to build open standards for multi-chip LLM inference.
Qwen Team has released **QwQ-32B-Preview**, an experimental model with enhanced deep reasoning capabilities in mathematics and coding.
Even in languages with a garbage collector, memory leaks can occur through things like global caches, and Replit discovered its agent's memory issues this way.
This article analyzes step by step the design principles and necessity of RoPE, a core technology in modern LLMs.
Rox boosts sales team productivity by 50% through an AI agent swarm powered by OpenAI models.
Banksalad's QA team unveiled a method for automatically generating complex MyData test data using LLM and prompt engineering.
Qwen2.5-Turbo has expanded its context length to 1 million tokens and significantly improved inference speed and cost efficiency.
Gartner stated that FuriosaAI's power-efficient chips deserve attention for reducing operational costs when adopting AI hardware.
Kakao has unveiled the 'Kanana' family of AI models specialized for Korean language processing, demonstrating overwhelming performance in Korean compared to global SOTA models and already applied to services such as KakaoTalk.
Kakao's Kanana Essence, trained on 3T tokens, achieved higher benchmark averages than llama-3.1-8b, while Kanana Nano maximized efficiency through pruning and distillation techniques.
The Kanana LLM, developed through SFT, DPO, and model merging, demonstrated superior performance compared to similar-sized models on Korean benchmarks such as LogicKor.
The Qwen2.5-Coder series, featuring GPT-4o-level coding performance, has been released as open source in a variety of model sizes.