# forck.live — 6–12 March 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-W10, 6–12 March 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-W10 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Runway — Gen-2: Generate novel videos with text, images or video clips - Company: Runway (runway.com) - Announced: 2023-03-11T10:10:10.110000+00:00 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://runway.com/research/gen-2 - Record: https://forck.live/items/7939-gen-2-generate-novel-videos-with-text-images-or-video-clips - Subject: Gen / Aleph Runway released Gen-2, a model that can generate novel videos from text prompts, images, or video clips. It offers multiple modes including text-to-video, image-to-video, and stylization. ### Hugging Face — Multivariate Probabilistic Time Series Forecasting with Informer - Company: Hugging Face (huggingface.co) - Announced: 2023-03-10 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/informer - Record: https://forck.live/items/2085-multivariate-probabilistic-time-series-forecasting-with-informer - Subject: Platform The Informer model for multivariate probabilistic time series forecasting, explains its improvements over the vanilla Transformer (ProbSparse attention and Distilling), and demonstrates how to use it in the 🤗 Transformers library. ### Hugging Face — Fine-tuning 20B LLMs with RLHF on a 24GB consumer GPU - Company: Hugging Face (huggingface.co) - Announced: 2023-03-09 - Category: developer-tool-release - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/trl-peft - Record: https://forck.live/items/2086-fine-tuning-20b-llms-with-rlhf-on-a-24gb-consumer-gpu - Subject: Platform Hugging Face announces the official integration of TRL with PEFT, enabling fine-tuning of large language models (up to 20B parameters) with RLHF on a single 24GB consumer GPU using adapters and 8-bit quantization. ### Replit — BerriAI—The Y Combinator company that brings LLM products to market quickly with Replit - Company: Replit (replit.com) - Announced: 2023-03-06T16:21:40+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replit.com/blog/berriai - Record: https://forck.live/items/18126-berriai-the-y-combinator-company-that-brings-llm-products-to-market-quickly - Subject: Replit Agent BerriAI, a Y Combinator-backed startup, uses Replit as a core development tool to prototype, collaborate, and host LLM applications. The company builds ChatGPT apps that connect user data to an LLM, enabling rapid development and deployment. BerriAI has created templates for various use cases including customer support bots and documentation Q&A. ### LG AI Research — EXAONE Lab, Dreaming of Super-Giant AI to Advance the World - Company: LG AI Research (lgresearch.ai) - Announced: 2023-03-06 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=309 - Record: https://forck.live/items/4905-exaone-lab-dreaming-of-super-giant-ai-to-advance-the-world - Subject: EXAONE LG AI Research has established the EXAONE Lab to accelerate development of its super-giant AI model EXAONE, which understands both language and images. ### Hugging Face — New ViT and ALIGN Models From Kakao Brain - Company: Hugging Face (huggingface.co) - Announced: 2023-03-06 - Category: open-weight-release - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/vit-align - Record: https://forck.live/items/2087-new-vit-and-align-models-from-kakao-brain - Subject: Platform Kakao Brain and Hugging Face release open-source ViT and ALIGN models trained on the new COYO dataset of 700 million image-text pairs, marking the first public open-source release of the ALIGN model and the first ViT and ALIGN models trained on an open dataset.