# forck.live — 19–25 December 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-W51, 19–25 December 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-W51 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Alibaba — Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese - Company: Alibaba (alibaba.com) - Announced: 2022-12-24T06:54:19+00:00 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://qwenlm.github.io/blog/chinese-clip/ - Record: https://forck.live/items/2324-chinese-clip-contrastive-vision-language-pretraining-in-chinese - Subject: Qwen Alibaba announced the launch of a project for Chinese CLIP, a contrastive vision-language pretraining model specifically for Chinese, to promote Chinese multimodal representation learning. ### OpenAI — The power of continuous learning - Company: OpenAI (openai.com) - Announced: 2022-12-23T08:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/the-power-of-continuous-learning - Record: https://forck.live/items/884-the-power-of-continuous-learning - Subject: GPT / ChatGPT / API Lilian Weng works on Applied AI Research at OpenAI. ### Replit — Redesigning Cycles Transaction History - Company: Replit (replit.com) - Announced: 2022-12-23 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/cycles-transaction-history - Record: https://forck.live/items/18047-redesigning-cycles-transaction-history - Subject: Replit Agent Earlier this year we introduced Cycles to the platform, a virtual token that can be earned or purchased to add compute power and functionality for your Repls. Hundreds of thousands of cycles transactions have been made to date. And now more Replit community members are able to earn cycles of their own through bounties! In the past, Replit transactions looked something like this: With this view, although simple, it’s hard to know where exactly your cycles are going, whether it's spending on powerups, purchasing, or sending to other replit community members via tips and bounties. There’s also nothing actionable you can do with this information. Our initial focus for the first iteration was purely around cycles functionalty. Now there are even more things you can use cycles for, we're revisiting the experience to focus on helping you find information on where you spent or earned cycles on Replit! As a goal of my internship, I wanted to work on a more complex engineering problem and learn more about the cycles team, which is why I took on this project! The goals for this project are to: Make the cycles transaction page extensible for future cycles products we build on replit … ### LG AI Research — [NeurIPS 2022] Transformers meet Stochastic Block Models: Attention with Data-Adaptive Sparsity and Cost - Company: LG AI Research (lgresearch.ai) - Announced: 2022-12-23 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://www.lgresearch.ai/blog/view?seq=291 - Record: https://forck.live/items/4912-neurips-2022-transformers-meet-stochastic-block-models-attention-with-data - Subject: EXAONE Proposes SBM-Transformer, a Transformer model that uses a Stochastic Block Model to data-adaptively choose attention sparsity and computational cost, enabling linear complexity in the number of sampled edges while preserving universal approximability. ### Replit — Custom Themes - Company: Replit (replit.com) - Announced: 2022-12-22 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/themes - Record: https://forck.live/items/18229-custom-themes - Subject: Replit Agent Today we're super excited to announce the launch of one of our most requested features to date. For almost as long as Replit has been around, people in the community have been asking for a way to create and share their very own Custom Themes! Well now you can. With Custom Themes, you can truly make Replit yours by styling everything from the app's colors, both in the IDE and across the site, to the syntax highlighting in your code. Create To get started, simply visit the themes section under your account page and hit "Create Theme". From there, give your new theme a title, add a description, and select a color scheme (light or dark). Once your theme is created, you'll be dropped into the Themes Editor. In the Themes Editor, you can change the global tokens, which include the background, foreground, primary, and outline colors that you'll see throughout the entire site. Just below that is the syntax higlighting editor where you can configure how various tokens appear in your code. These will apply in the IDE as well as everywhere else you might see a code editor. … ### LG AI Research — [NeurIPS 2022] HLA-Peptide Binding Prediction Model for the Personalized Treatment of Cancer - Company: LG AI Research (lgresearch.ai) - Announced: 2022-12-21 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/blog/view?seq=290 - Record: https://forck.live/items/4913-neurips-2022-hla-peptide-binding-prediction-model-for-the-personalized - Subject: EXAONE LG AI Research proposed a conditional attention mechanism for HLA-peptide binding affinity prediction, aimed at personalized cancer vaccines, and published a paper at the NeurIPS 2022 Workshop on Machine Learning and Structural Biology. ### Hugging Face — Zero-shot image segmentation with CLIPSeg - Company: Hugging Face (huggingface.co) - Announced: 2022-12-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/clipseg-zero-shot - Record: https://forck.live/items/2117-zero-shot-image-segmentation-with-clipseg - Subject: Platform To use the CLIPSeg zero-shot image segmentation model with the Hugging Face Transformers library. ### Hugging Face — Model Cards - Company: Hugging Face (huggingface.co) - Announced: 2022-12-20 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/model-cards - Record: https://forck.live/items/2118-model-cards - Subject: Platform Hugging Face announces the release of a model card creation tool, an updated template, an annotated template, a user study, and a landscape analysis, aiming to improve ML documentation accessibility and inclusion.