# forck.live — 17–23 June 2024 > 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 2024-W25, 17–23 June 2024, 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/2024-W25 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### OpenAI — OpenAI acquires Rockset - Company: OpenAI (openai.com) - Announced: 2024-06-21T08:00:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/openai-acquires-rockset - Record: https://forck.live/items/738-openai-acquires-rockset - Subject: GPT / ChatGPT / API OpenAI has acquired Rockset. ### LG AI Research — The New Hwadamsup: Image Understanding & Generation through LMM - Company: LG AI Research (lgresearch.ai) - Announced: 2024-06-21 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/blog/view?seq=442 - Record: https://forck.live/items/4848-the-new-hwadamsup-image-understanding-generation-through-lmm - Subject: EXAONE LG AI Research introduces the New Hwadamsup project, a human-AI collaboration for media art, and describes their EXAONE LMM for image understanding and EXAONE image generation model, including a customized model for Hwadam botanic garden. ### Replicate — Replicate Intelligence #5 - Company: Replicate (replicate.com) - Announced: 2024-06-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replicate.com/blog/replicate-intelligence-2024-06-21 - Record: https://forck.live/items/2527-replicate-intelligence-5 - Subject: Platform This is a weekly newsletter from Replicate covering open-source AI models, tools, and research. It mentions DeepSeek-Coder-V2, an open-weights code model that briefly topped coding benchmarks, and PowerInfer-2, a fast inference engine for phones along with TurboSparse-Mistral-7B and TurboSparse-Mixtral-47B models. It also discusses a research essay on AI search and tree search methods, and announces a new support bot feature in Discord. ### OpenAI — Empowering defenders through our Cybersecurity Grant Program - Company: OpenAI (openai.com) - Announced: 2024-06-20T10:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/empowering-defenders-through-our-cybersecurity-grant-program - Record: https://forck.live/items/739-empowering-defenders-through-our-cybersecurity-grant-program - Subject: GPT / ChatGPT / API Innovative research and AI integration in cybersecurity through a grant program. ### LG AI Research — Taking Steps to Secure Future Talent at Global AI Conference - Company: LG AI Research (lgresearch.ai) - Announced: 2024-06-20 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=440 - Record: https://forck.live/items/11025-taking-steps-to-secure-future-talent-at-global-ai-conference - Subject: EXAONE LG AI Research attended CVPR 2024 in Seattle to recruit AI talent, hosted an LG AI Day networking event for over 100 graduate students, published 11 papers across LG affiliates, and organized a zero-shot image captioning workshop. The company also displayed media art created with its EXAONE generative multimodal AI, which was trained on 350 million licensed image-text pairs plus 6,700 garden images. ### LG AI Research — Taking Steps to Secure Future Talent at Global AI Conference - Company: LG AI Research (lgresearch.ai) - Announced: 2024-06-20 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=440 - Record: https://forck.live/items/4849-taking-steps-to-secure-future-talent-at-global-ai-conference - Subject: EXAONE LG AI Research attended CVPR 2024 to recruit AI talent, hosted LG AI Day, published papers, organized a zero-shot image captioning workshop, and displayed media art created with its generative multimodal AI EXAONE. ### Hugging Face — Data Is Better Together: A Look Back and Forward - Company: Hugging Face (huggingface.co) - Announced: 2024-06-20 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/dibt - Record: https://forck.live/items/1868-data-is-better-together-a-look-back-and-forward - Subject: Platform Hugging Face and Argilla reflect on the Data Is Better Together initiative, highlighting community dataset-building efforts such as the prompt ranking project and the Multilingual Prompt Evaluation Project, and outline future plans for community and cookbook efforts. ### OpenAI — Consistency Models - Company: OpenAI (openai.com) - Announced: 2024-06-20 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/consistency-models - Record: https://forck.live/items/742-consistency-models - Subject: GPT / ChatGPT / API A limitation of diffusion models: their iterative sampling process causes slow generation. ### OpenAI — A Holistic Approach to Undesired Content Detection in the Real World - Company: OpenAI (openai.com) - Announced: 2024-06-20 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/a-holistic-approach-to-undesired-content-detection-in-the-real-world - Record: https://forck.live/items/741-a-holistic-approach-to-undesired-content-detection-in-the-real-world - Subject: GPT / ChatGPT / API OpenAI presents a holistic approach to building a natural language classification system for real-world content moderation. ### OpenAI — Improved Techniques for Training Consistency Models - Company: OpenAI (openai.com) - Announced: 2024-06-20 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/improved-techniques-for-training-consistency-models - Record: https://forck.live/items/740-improved-techniques-for-training-consistency-models - Subject: GPT / ChatGPT / API OpenAI describes consistency models as a new family of generative models capable of one-step high-quality sampling without adversarial training. ### Supabase — Self-host Maps with Protomaps and Supabase Storage - Company: Supabase (supabase.com) - Announced: 2024-06-19 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://supabase.com/blog/self-host-maps-storage-protomaps - Record: https://forck.live/items/21423-self-host-maps-with-protomaps-and-supabase-storage - Subject: Supabase Supabase published a tutorial on self-hosting maps using Protomaps, an open source map of the world deployable as a single static file on Supabase Storage. The guide covers extracting an area into a PMTiles file, uploading it to Supabase Storage, rendering the map with MapLibre GL, and restricting access using Supabase Edge Functions. ### Hugging Face — Going multimodal: How Prezi is leveraging the Hub and the Expert Support Program to accelerate their ML roadmap - Company: Hugging Face (huggingface.co) - Announced: 2024-06-19 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/prezi-case-study - Record: https://forck.live/items/1869-going-multimodal-how-prezi-is-leveraging-the-hub-and-the-expert-support - Subject: Platform Prezi case study on using Hugging Face Expert Support Program to integrate multimodal open-source models into their presentation AI. ### Vercel — Introducing Vercel AI SDK 3.2 - Company: Vercel (vercel.com) - Announced: 2024-06-18T13:00:00+00:00 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/introducing-vercel-ai-sdk-3-2 - Record: https://forck.live/items/20971-introducing-vercel-ai-sdk-3-2 - Subject: v0 / AI SDK / AI Gateway Vercel released AI SDK 3.2, which extends generateText and streamText for multi-step agentic workflows, adds embeddings support for embeddings with OpenAI, Mistral, Azure OpenAI, and Ollama providers, and improves observability with onFinish callback for token usage and errors. ### OpenAI — Surging developer productivity with custom GPTs - Company: OpenAI (openai.com) - Announced: 2024-06-18T08:45:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/paf - Record: https://forck.live/items/743-surging-developer-productivity-with-custom-gpts - Subject: GPT / ChatGPT / API Paf adopted ChatGPT Enterprise company-wide, with engineers using custom GPTs daily to speed up routine development tasks. Additionally, 70% of Paf employees actively use ChatGPT Enterprise across various business teams. The ChatGPT Enterprise is also integrated into the grit:lab coding academy for training software developers. ### Supabase — Cal.com launches Expert Marketplace built with Next.js and Supabase. - Company: Supabase (supabase.com) - Announced: 2024-06-18 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://supabase.com/blog/calcom-platform-starter-kit-nextjs-supabase - Record: https://forck.live/items/21424-cal-com-launches-expert-marketplace-built-with-next-js-and-supabase - Subject: Supabase Cal.com and Supabase collaborated on a platform starter kit for building a marketplace with scheduling, using Next.js, Prisma, and Supabase for Postgres, storage, and authentication. The kit is open-source and includes features like file uploads, signed URLs, and Vercel integration. ### Upstage — Solar LLM with Predibase: The best LLM for fine-tuning that beats GPT-4 - Company: Upstage (upstage.ai) - Announced: 2024-06-18 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.upstage.ai/blog/en/solar-llm-with-predibase-the-best-llm-for-fine-tuning - Record: https://forck.live/items/18481-solar-llm-with-predibase-the-best-llm-for-fine-tuning-that-beats-gpt-4 - Subject: Solar Pro / Solar Mini For organizations with domain-specific data and use cases, fine-tuning is one of the most performant and cost-effective ways to tailor LLMs for production applications. ### Factory AI — Factory Raises $15M Series A from Sequoia - Company: Factory AI (factory.com) - Announced: 2024-06-18 - Category: funding-company-news - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://factory.com/news/series-a-announcement - Record: https://forck.live/items/17869-factory-raises-15m-series-a-from-sequoia - Subject: Droid Factory announced a $15M Series A led by Sequoia Capital, with participation from Lux Capital and Mantis VC, bringing total funding to over $20M and valuing the company at $120M. The funding will be used to further the company's mission, grow the team, and improve its Droids. Factory also reported that its Code Droid set a new state-of-the-art on the SWE-bench coding benchmark. ### Factory AI — Code Droid: A Technical Report - Company: Factory AI (factory.com) - Announced: 2024-06-18 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://factory.com/news/code-droid-technical-report - Record: https://forck.live/items/17822-code-droid-a-technical-report - Subject: Droid Factory published a technical report on Code Droid, an autonomous system for software engineering tasks. The report states that Code Droid achieved 19.27% on SWE-bench Full and 31.67% on SWE-bench Lite. The system uses techniques including planning and task decomposition, tool integration, HyperCode and ByteRank for codebase understanding, and multi-model sampling. ### LG AI Research — The First Summer Internship 2024 Has Begun! - Company: LG AI Research (lgresearch.ai) - Announced: 2024-06-18 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=438 - Record: https://forck.live/items/11026-the-first-summer-internship-2024-has-begun - Subject: EXAONE LG AI Research has launched its first Summer Internship 2024 for overseas master's and doctoral students. Six interns were selected after document screening, coding tests, and technical interviews. The three-month program (June–August) includes onboarding, orientation, mentorship, and research projects at LG AI Research facilities. ### LG AI Research — The First Summer Internship 2024 Has Begun! - Company: LG AI Research (lgresearch.ai) - Announced: 2024-06-18 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=438 - Record: https://forck.live/items/4850-the-first-summer-internship-2024-has-begun - Subject: EXAONE LG AI Research announces the start of its first "Summer Internship" program, an overseas recruitment initiative for master's and doctoral students from foreign universities, with six interns selected for 2024. ### Replicate — Run Stable Diffusion 3 on your Apple Silicon Mac - Company: Replicate (replicate.com) - Announced: 2024-06-18 - Category: developer-tool-release - Coverage: not counted - Announcement: no - Group: routine - Source: https://replicate.com/blog/run-stable-diffusion-3-on-apple-silicon-mac - Record: https://forck.live/items/2529-run-stable-diffusion-3-on-your-apple-silicon-mac - Subject: Platform To run Stable Diffusion 3 locally on Apple Silicon Macs, introducing DiffusionKit as a CLI tool for easier use. ### Replicate — How to get the best results from Stable Diffusion 3 - Company: Replicate (replicate.com) - Announced: 2024-06-18 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replicate.com/blog/get-the-best-from-stable-diffusion-3 - Record: https://forck.live/items/2528-how-to-get-the-best-results-from-stable-diffusion-3 - Subject: Platform This is a how-to guide for getting the best results from Stable Diffusion 3 Medium, covering version selection, prompting techniques, and configurations. It does not announce a new product, model release, or API. ### Hugging Face — BigCodeBench: The Next Generation of HumanEval - Company: Hugging Face (huggingface.co) - Announced: 2024-06-18 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/leaderboard-bigcodebench - Record: https://forck.live/items/1870-bigcodebench-the-next-generation-of-humaneval - Subject: Platform Hugging Face announces the release of BigCodeBench, a new benchmark for evaluating LLMs on practical and challenging programming tasks without contamination, containing 1,140 function-level tasks from 139 libraries. ### Runway — Runway Gen-3 Alpha: AI Video Generation with Temporal Control - Company: Runway (runway.com) - Announced: 2024-06-17T10:10:10.110000+00:00 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://runway.com/research/introducing-gen-3-alpha - Record: https://forck.live/items/7926-runway-gen-3-alpha-ai-video-generation-with-temporal-control - Subject: Gen / Aleph Runway released Gen-3 Alpha, a new AI foundation model for video generation with temporal control. It is trained jointly on videos and images and offers improvements in fidelity, consistency, and motion over Gen-2. The model will power Runway's text-to-video, image-to-video, and text-to-image tools, and includes new safeguards. ### OpenAI — Using GPT-4o reasoning to transform cancer care - Company: OpenAI (openai.com) - Announced: 2024-06-17T04:15:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/color-health - Record: https://forck.live/items/745-using-gpt-4o-reasoning-to-transform-cancer-care - Subject: GPT / ChatGPT / API OpenAI and Color Health announce the Cancer Copilot application that uses GPT-4o to identify missing diagnostics and create tailored workup plans for cancer care.