# forck.live — 19–25 February 2024 > AI lab announcements, news reports and podcasts. forck.live follows what companies announce about their own products. Press may corroborate an official story; it never leads and never appears as an item of its own. A missing date stays missing. A title is the publisher's words — where a company does not publish in English, an English rendering is listed separately as `Title in English` and named as this site's translation, never as theirs. This file is one issue of the weekly archive: every confirmed first-party announcement of ISO week 2024-W08, 19–25 February 2024, newest first. The window never moves, so this URL always names the same seven days — but a post published inside them and detected later is added when it arrives, so a closed week can still gain entries. Every entry is one first-party announcement, and carries the same eight fields in this order before any others: Company · Announced · Category · Coverage · Announcement · Group · Source · Record `Source` is the company's own page the announcement was traced to and `Record` is this site's page for it. Both are safe to cite. `Coverage` is a number of outlets, or `not counted` — which means this record reached no count of its own, never that nobody wrote about it. `Category` is this tracker's own analysis, and is authoritative wherever it is stated. `not stated` means either that nothing has analysed the post yet or that the analysis placed it outside this taxonomy; neither is a verdict about the announcement. `Announcement` is whether the post is news of something the company did, rather than a how-to, a customer story, a changelog listing or a staffing note. `Group` is which of this record's groups the entry falls in, carried on the entry itself so it holds where the groups are headings and where they are not. One of: `models`, `covered`, `announcements`, `routine`. Anything after `Record` is held detail that only some announcements carry. - This week's page: https://forck.live/week/2024-W08 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Hugging Face — 🪆 Introduction to Matryoshka Embedding Models - Company: Hugging Face (huggingface.co) - Announced: 2024-02-23 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/matryoshka - Record: https://forck.live/items/1937-introduction-to-matryoshka-embedding-models - Subject: Platform Hugging Face introduces Matryoshka embedding models, which allow embeddings to be truncated to smaller dimensions while retaining performance, improving efficiency for downstream tasks. ### Hugging Face — Introducing the Red-Teaming Resistance Leaderboard - Company: Hugging Face (huggingface.co) - Announced: 2024-02-23 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/leaderboard-haizelab - Record: https://forck.live/items/1936-introducing-the-red-teaming-resistance-leaderboard - Subject: Platform Hugging Face and Haize Labs announced the Red-Teaming Resistance Leaderboard, a benchmark to evaluate LLM robustness against adversarial prompts. ### Hugging Face — Fine-Tuning Gemma Models in Hugging Face - Company: Hugging Face (huggingface.co) - Announced: 2024-02-23 - Category: developer-tool-release - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/gemma-peft - Record: https://forck.live/items/1935-fine-tuning-gemma-models-in-hugging-face - Subject: Platform Hugging Face announces support for fine-tuning Google DeepMind's Gemma models using Parameter Efficient Fine-Tuning (PEFT) techniques, specifically LoRA and QLoRA, with the Hugging Face Transformers and PEFT libraries on GPUs and Cloud TPUs. ### Replit — Experiment: Figma to Replit Plugin - Company: Replit (replit.com) - Announced: 2024-02-22T14:00:00+00:00 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/figma-to-replit - Record: https://forck.live/items/17954-experiment-figma-to-replit-plugin - Subject: Replit Agent Replit released an experimental Figma plugin that generates HTML, CSS, and React code from Figma designs, allowing users to create a Repl directly from a design and share a static React app. The plugin supports one-click code generation and export to Replit, but has limitations such as no image fills, gradients, shadows, or component sets. It was a Hack Week project and may not be maintained in the future. ### Upstage — Introducing Solar API Beta - Company: Upstage (upstage.ai) - Announced: 2024-02-22 - Category: availability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.upstage.ai/news/solar-api-beta - Record: https://forck.live/items/18447-introducing-solar-api-beta - Subject: Solar Pro / Solar Mini Upstage released the Solar API Beta, offering free access to its Solar LLM until March 31st. The model, which topped the Open LLM Leaderboard, achieves performance comparable to GPT-3.5 with fewer parameters and 2.5x faster speeds. The API also features English-Korean translation surpassing GPT-4 and DeepL. ### Vercel — Effortless high availability for dynamic frontends - Company: Vercel (vercel.com) - Announced: 2024-02-21T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/effortless-high-availability-for-dynamic-frontends - Record: https://forck.live/items/21008-effortless-high-availability-for-dynamic-frontends - Subject: v0 / AI SDK / AI Gateway Vercel's Frontend Cloud provides high availability through global anycast routing, multi-AZ and multi-region redundancy with Function Failover, and Edge Config for backend failover without redeployment. ### Supabase — Automating performance tests - Company: Supabase (supabase.com) - Announced: 2024-02-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://supabase.com/blog/automating-performance-tests - Record: https://forck.live/items/21446-automating-performance-tests - Subject: Supabase Performance testing is one of the non-functional testing types that evaluates a system's compliance with its performance requirements. It reveals if your app can handle user load, unexpected spikes, or recover from stressful workloads. ### LG AI Research — LG AI Research in the Global Forum on Ethics of AI - Company: LG AI Research (lgresearch.ai) - Announced: 2024-02-21 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/news/view?seq=410 - Record: https://forck.live/items/11033-lg-ai-research-in-the-global-forum-on-ethics-of-ai - Subject: EXAONE LG AI Research participated in the 2nd UNESCO Global Forum on Ethics of AI held in Kranj, Slovenia in February 2024, as the sole South Korean company in attendance. The forum's theme was 'Changing the Landscape of AI Governance' and it included over 500 participants from 67 countries. Discussions focused on inclusive AI governance, risks such as inequality and job displacement, and the need for diverse policy approaches between developed and developing nations. ### LG AI Research — LG AI Research in the Global Forum on Ethics of AI - Company: LG AI Research (lgresearch.ai) - Announced: 2024-02-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=410 - Record: https://forck.live/items/4864-lg-ai-research-in-the-global-forum-on-ethics-of-ai - Subject: EXAONE LG AI Research reports on its participation in the 2nd UNESCO Global Forum on Ethics of AI, covering discussions on global AI governance, inequality, and inclusivity. No product or model announcements are made. ### Hugging Face — Welcome Gemma - Google’s new open LLM - Company: Hugging Face (huggingface.co) - Announced: 2024-02-21 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/gemma - Record: https://forck.live/items/1938-welcome-gemma-google-s-new-open-llm - Subject: Platform Google released Gemma, a new family of open LLMs in two sizes (7B and 2B parameters), each with base and instruction-tuned variants, supporting 8K token context and available on Hugging Face Hub with integrations. ### Hugging Face — Introducing the Open Ko-LLM Leaderboard: Leading the Korean LLM Evaluation Ecosystem - Company: Hugging Face (huggingface.co) - Announced: 2024-02-20 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/leaderboard-upstage - Record: https://forck.live/items/1939-introducing-the-open-ko-llm-leaderboard-leading-the-korean-llm-evaluation - Subject: Platform Upstage and Hugging Face introduce the Open Ko-LLM Leaderboard, a new evaluation ecosystem for Korean LLMs with Korean-language datasets, private test sets to prevent contamination, and five evaluation tasks. The leaderboard has received over 1,000 model submissions from various organizations. ### Hugging Face — 🤗 PEFT welcomes new merging methods - Company: Hugging Face (huggingface.co) - Announced: 2024-02-19 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/peft_merging - Record: https://forck.live/items/1940-peft-welcomes-new-merging-methods - Subject: Platform Hugging Face announces new merging methods for LoRA adapters in their PEFT library.