# forck.live — 23–29 January 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-W04, 23–29 January 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-W04 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Replit — Replit x India (Part 2) - Company: Replit (replit.com) - Announced: 2023-01-27T23:56:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replit.com/blog/replit-x-india-part-2 - Record: https://forck.live/items/18028-replit-x-india-part-2 - Subject: Replit Agent Introducing the official launch of Replit India - Replit's first international launch! Thank you to the 2M+ developers from India who have signed up for Replit to learn to code, and host & deploy projects. Replit's mission is to bring the next billion coders online, and this is not possible without India. We have launched a plethora of products and initiatives for Indian developers to learn to code, earn Cycles anywhere, anytime and host & deploy their project on Replit 24/7. First some eye-popping stats on India and coding: India is our 2nd largest developer community and fastest growing market ~600K Repls are created every month from India CodeWithHarry from India is the most followed person on Replit This is what we have launched as part of Replit India! Code anywhere and anytime Dedicated Servers in India We just launched our first international server cluster in Mumbai. This will drastically reduce latency for our India-based users to less than 100ms. Coding on Replit will now be even smoother - try it out today! … ### LG AI Research — [ISMIR / NeurIPS 2022] Symbolic Music Motif Generation - Company: LG AI Research (lgresearch.ai) - Announced: 2023-01-27 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/blog/view?seq=297 - Record: https://forck.live/items/4910-ismir-neurips-2022-symbolic-music-motif-generation - Subject: EXAONE LG AI Research introduces two academic papers on symbolic music motif generation, presented at ISMIR 2022 and the NeurIPS 2022 Workshop. The research focuses on generating musical motifs—short 4- or 8-measure loops—using AI, including a loop detector based on anomaly detection and a VQ-VAE with LSTM for generation. The post describes the methodology and evaluation metrics. ### Hugging Face — Using LoRA for Efficient Stable Diffusion Fine-Tuning - Company: Hugging Face (huggingface.co) - Announced: 2023-01-26 - Category: developer-tool-release - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/lora - Record: https://forck.live/items/2106-using-lora-for-efficient-stable-diffusion-fine-tuning - Subject: Platform Hugging Face announces LoRA training support in their diffusers library, enabling efficient fine-tuning of Stable Diffusion with reduced compute and memory requirements, and significantly smaller trained weight files (~3 MB). ### Hugging Face — 2D Asset Generation: AI for Game Development #4 - Company: Hugging Face (huggingface.co) - Announced: 2023-01-26 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/ml-for-games-4 - Record: https://forck.live/items/2105-2d-asset-generation-ai-for-game-development-4 - Subject: Platform A tutorial blog post on using AI, specifically Stable Diffusion with Image2Image, to generate 2D game assets for a farming game as part of a series on AI for game development. ### Hugging Face — Optimum+ONNX Runtime - Easier, Faster training for your Hugging Face models - Company: Hugging Face (huggingface.co) - Announced: 2023-01-24 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/optimum-onnxruntime-training - Record: https://forck.live/items/2108-optimum-onnx-runtime-easier-faster-training-for-your-hugging-face-models - Subject: Platform Hugging Face and Microsoft's ONNX Runtime announce the integration of ONNX Runtime training support into the Optimum library, enabling up to 130% training speedup (with DeepSpeed ZeRO Stage 1) for popular Hugging Face Transformer models. The ORTTrainer API simplifies adoption with ease-of-use similar to the Transformers Trainer. ### Hugging Face — What Makes a Dialog Agent Useful? - Company: Hugging Face (huggingface.co) - Announced: 2023-01-24 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/dialog-agents - Record: https://forck.live/items/2107-what-makes-a-dialog-agent-useful - Subject: Platform This blog post surveys the techniques behind ChatGPT, including RLHF, IFT, CoT, and red teaming, and compares several dialog agents such as LaMDA, BlenderBot 3, Sparrow, ChatGPT/InstructGPT, and Assistant. It discusses instruction fine-tuning and safety in dialog agents. ### Replit — PostgreSQL Database Preview Now Available On Replit - Company: Replit (replit.com) - Announced: 2023-01-23T17:15:00+00:00 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/postgresql-db-launch - Record: https://forck.live/items/18036-postgresql-database-preview-now-available-on-replit - Subject: Replit Agent Databases and web apps go together like peanut butter and jelly. In a word, they're inseparable. And despite all the amazing innovations in NoSQL data stores, often a good old relational database is the most reliable tool for the job. We want to make it completely seamless to develop applications that need databases on Replit. Starting today, you can create and instantly begin to use PostgreSQL databases from within the Replit workspace. Here's how you use it. First, click the PostgreSQL icon in the Tools pane on the lower left hand side of the workspace. Or, you can open a new tab and select PostgreSQL from there. From this pane, you can provision a PostgreSQL database. Compared to most SQL offerings which can take several minutes, your database on Replit will be available within just a few seconds. For now we are only offering one size of database: 100 cycles per day for 10GB. The database has 1 dedicated CPU and 4 GB of RAM to process your queries. Once you've created the database, you can start using it! The DATABASE_URL environment variable will have your connection string. … ### OpenAI — OpenAI and Microsoft extend partnership - Company: OpenAI (openai.com) - Announced: 2023-01-23T08:00:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/openai-and-microsoft-extend-partnership - Record: https://forck.live/items/879-openai-and-microsoft-extend-partnership - Subject: GPT / ChatGPT / API OpenAI and Microsoft are extending their partnership.