# forck.live — 11–17 July 2022 > 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 2022-W28, 11–17 July 2022, 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/2022-W28 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Hugging Face — How to train your model dynamically using adversarial data - Company: Hugging Face (huggingface.co) - Announced: 2022-07-16 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/mnist-adversarial - Record: https://forck.live/items/2176-how-to-train-your-model-dynamically-using-adversarial-data - Subject: Platform The concept of dynamic adversarial data collection (DADC) and provides a step-by-step tutorial on how to implement it using an MNIST handwritten digit recognition model with Hugging Face Spaces and Gradio. ### Replit — Inspect your HTML/CSS/JS Repls with native DevTools - Company: Replit (replit.com) - Announced: 2022-07-15 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/devtools - Record: https://forck.live/items/18048-inspect-your-html-css-js-repls-with-native-devtools - Subject: Replit Agent Replit launched a native DevTools feature for HTML/CSS/JS repls, providing console and DOM inspection directly within the platform. The tools are built on the open-source Eruda library and use a custom injection approach to capture events before page scripts run, addressing limitations of browser DevTools on Replit. ### OpenAI — DALL·E 2: Extending creativity - Company: OpenAI (openai.com) - Announced: 2022-07-14T07:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/dall-e-2-extending-creativity - Record: https://forck.live/items/900-dall-e-2-extending-creativity - Subject: GPT / ChatGPT / API OpenAI reports that over 3,000 artists from 118 countries have used DALL·E 2 during its research preview, helping to identify new applications and inform feature decisions. ### Hugging Face — The Technology Behind BLOOM Training - Company: Hugging Face (huggingface.co) - Announced: 2022-07-14 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/bloom-megatron-deepspeed - Record: https://forck.live/items/2177-the-technology-behind-bloom-training - Subject: Platform The technology and engineering behind training the 176B parameter BLOOM model, including hardware, software stack, and parallelization techniques. ### Replit — Leaky UIs - Company: Replit (replit.com) - Announced: 2022-07-13T00:30:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replit.com/blog/leaky-uis - Record: https://forck.live/items/18056-leaky-uis - Subject: Replit Agent Replit's workspace layout system uses a binary tree data structure, which causes unintuitive resizing behavior when panes are grouped. The company plans to switch to an n-ary tree, where panes that appear side-by-side are siblings, making resizing affect only the directly adjacent panes. ### Hugging Face — Building a Playlist Generator with Sentence Transformers - Company: Hugging Face (huggingface.co) - Announced: 2022-07-13 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/playlist-generator - Record: https://forck.live/items/2178-building-a-playlist-generator-with-sentence-transformers - Subject: Platform To build a playlist generator using Sentence Transformers and Gradio, focusing on generating embeddings and creating a multi-step demo. ### Hugging Face — Introducing The World's Largest Open Multilingual Language Model: BLOOM - Company: Hugging Face (huggingface.co) - Announced: 2022-07-12 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/bloom - Record: https://forck.live/items/2179-introducing-the-world-s-largest-open-multilingual-language-model-bloom - Subject: Platform Hugging Face releases BLOOM, a 176-billion-parameter open multilingual language model trained in complete transparency, covering 46 natural languages and 13 programming languages, under a Responsible AI License, as the result of a large collaboration of over 1000 researchers.