# forck.live — 15–21 May 2023 > 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 2023-W20, 15–21 May 2023, 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/2023-W20 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Replit — May 18 Replit downtime - Company: Replit (replit.com) - Announced: 2023-05-20T00:30:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replit.com/blog/may-18-replit-downtime - Record: https://forck.live/items/18185-may-18-replit-downtime - Subject: Replit Agent Replit experienced a two-hour outage on May 18 caused by a deadlock from a leaked read-lock and a configuration change that triggered a write-lock. The rollback process was delayed, and the autoscaler reduced capacity due to missing signals. The root cause was fixed, and Replit is implementing staggered deployments, improved health checks, and disaster recovery training. ### OpenAI — Introducing the ChatGPT app for iOS - Company: OpenAI (openai.com) - Announced: 2023-05-18T07:00:00+00:00 - Category: product-launch - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/introducing-the-chatgpt-app-for-ios - Record: https://forck.live/items/861-introducing-the-chatgpt-app-for-ios - Subject: GPT / ChatGPT / API OpenAI announces the ChatGPT app for iOS, which syncs conversations, supports voice input, and includes the latest model improvements. ### Replicate — Status page - Company: Replicate (replicate.com) - Announced: 2023-05-18 - Category: infrastructure-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replicate.com/blog/new-status-page - Record: https://forck.live/items/2567-status-page - Subject: Platform Replicate launched a new status page at replicatestatus.com to provide real-time updates on service health and plans to publish detailed incident reports after major outages. ### Hugging Face — Smaller is better: Q8-Chat, an efficient generative AI experience on Xeon - Company: Hugging Face (huggingface.co) - Announced: 2023-05-16 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/generative-ai-models-on-intel-cpu - Record: https://forck.live/items/2060-smaller-is-better-q8-chat-an-efficient-generative-ai-experience-on-xeon - Subject: Platform Intel and Hugging Face show that SmoothQuant quantization reduces model size by ~2x and enables efficient LLM inference on Intel CPUs, as demonstrated with models like OPT, LLaMA, Alpaca, Vicuna, BloomZ, and MPT-7B-chat. ### Hugging Face — Large-scale Near-deduplication Behind BigCode - Company: Hugging Face (huggingface.co) - Announced: 2023-05-16 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/dedup - Record: https://forck.live/items/2059-large-scale-near-deduplication-behind-bigcode - Subject: Platform The large-scale near-deduplication approach used in the BigCode project, discussing motivations, methods, and results. ### Replit — Making Git Good - Company: Replit (replit.com) - Announced: 2023-05-15 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/git-good - Record: https://forck.live/items/18013-making-git-good - Subject: Replit Agent There has been a new Git pane in Repls for a little while now. This new Git UI you see is a part of a complete rewrite of everything Git-related on Replit: both engineering and design. It is the start of our journey into more deeply integrating Git into Replit, with much more to come. Let's talk about it! Design challenges Making Git more accessible Git can be hard to use. Whether you’re a professional developer or a beginner, it’s incredibly easy to make simple mistakes that seem impossible to fix and can suck hours out of your productivity. Ultimately, dealing with bugs and misunderstandings like these sink your attention into learning obscure tools and distract you from what you're actually trying to build. That being said, we didn't try to do things like rename terms or introduce new concepts. Everything you see in the new UI is "regular git" except we really tried to tie a few concepts together visually and simplify workflows (multiple steps you'd usually take in a row while working on a project). For example, grouping commits, staging, and .gitignore data together. … ### Hugging Face — Introducing RWKV - An RNN with the advantages of a transformer - Company: Hugging Face (huggingface.co) - Announced: 2023-05-15 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/rwkv - Record: https://forck.live/items/2063-introducing-rwkv-an-rnn-with-the-advantages-of-a-transformer - Subject: Platform Hugging Face announces the integration of the RWKV architecture into its transformers library. RWKV combines the strengths of RNNs and transformers, offering efficient long-context processing. The blog details the architecture and mentions RWKV-4 models with a context length of 8192 tokens. ### Hugging Face — Run a Chatgpt-like Chatbot on a Single GPU with ROCm - Company: Hugging Face (huggingface.co) - Announced: 2023-05-15 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/chatbot-amd-gpu - Record: https://forck.live/items/2062-run-a-chatgpt-like-chatbot-on-a-single-gpu-with-rocm - Subject: Platform Running the Vicuna 13B open-source chatbot model on a single AMD GPU using ROCm, including quantization with GPTQ to reduce memory footprint. ### Hugging Face — Hugging Face Selected for the French Data Protection Agency Enhanced Support Program - Company: Hugging Face (huggingface.co) - Announced: 2023-05-15 - Category: funding-company-news - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/cnil - Record: https://forck.live/items/2061-hugging-face-selected-for-the-french-data-protection-agency-enhanced-support - Subject: Platform Hugging Face was selected by the French Data Protection Authority (CNIL) for its Enhanced Support program, which aims to assist the company with GDPR compliance and data protection in AI.