# forck.live — 8–14 July 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-W28, 8–14 July 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-W28 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Replicate — Replicate Intelligence #7 - Company: Replicate (replicate.com) - Announced: 2024-07-12 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://replicate.com/blog/replicate-intelligence-2024-07-12 - Record: https://forck.live/items/2525-replicate-intelligence-7 - Subject: Platform Fal.ai released AuraFlow, a 6.8 billion parameter open-source text-to-image model with optimized architecture and novel training approach. ### Supabase — Simplifying Time-Based Queries with Range Columns - Company: Supabase (supabase.com) - Announced: 2024-07-11 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://supabase.com/blog/range-columns - Record: https://forck.live/items/21420-simplifying-time-based-queries-with-range-columns - Subject: Supabase Supabase's guide explains how PostgreSQL range types (tstzrange, daterange, int4range) can replace separate start/end columns to simplify overlap queries and add exclusion constraints that prevent conflicting reservations, including per-table constraints using the btree_gist extension. ### Supabase — Supabase Security Suite - Company: Supabase (supabase.com) - Announced: 2024-07-11 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://supabase.com/blog/hardening-supabase - Record: https://forck.live/items/21419-supabase-security-suite - Subject: Supabase Supabase introduced several security features and guides, including a Security Advisor with one-click fixes, weekly security email notifications, the ability to disable the Data API at project creation or in settings, API hardening with schema switching, column-level security management, user impersonation in the dashboard, an RLS AI Assistant powered by GPT-4o, and network restrictions for database access. ### Hugging Face — How NuminaMath Won the 1st AIMO Progress Prize - Company: Hugging Face (huggingface.co) - Announced: 2024-07-11 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/winning-aimo-progress-prize - Record: https://forck.live/items/1855-how-numinamath-won-the-1st-aimo-progress-prize - Subject: Platform The NuminaMath 7B TIR model, fine-tuned by Numina and Hugging Face, won the 1st AIMO Progress Prize by solving 29 out of 50 math problems, and details the training recipe and techniques used. ### OpenAI — OpenAI and Los Alamos National Laboratory announce research partnership - Company: OpenAI (openai.com) - Announced: 2024-07-10T06:30:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/openai-and-los-alamos-national-laboratory-work-together - Record: https://forck.live/items/735-openai-and-los-alamos-national-laboratory-announce-research-partnership - Subject: GPT / ChatGPT / API OpenAI and Los Alamos National Laboratory announced a research partnership to develop safety evaluations for assessing biological risks from frontier AI models. ### Hugging Face — Preference Optimization for Vision Language Models - Company: Hugging Face (huggingface.co) - Announced: 2024-07-10 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/dpo_vlm - Record: https://forck.live/items/1858-preference-optimization-for-vision-language-models - Subject: Platform Hugging Face announces that the TRL library now supports direct preference optimization (DPO) for vision language models (VLMs), including Idefics2-8b, Llava 1.5, and PaliGemma. ### Hugging Face — Experimenting with Automatic PII Detection on the Hub using Presidio - Company: Hugging Face (huggingface.co) - Announced: 2024-07-10 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/presidio-pii-detection - Record: https://forck.live/items/1857-experimenting-with-automatic-pii-detection-on-the-hub-using-presidio - Subject: Platform Hugging Face is experimenting with a new feature on the Dataset Hub that uses Presidio, an open-source PII detection tool, to automatically detect and report the presence of personally identifying information (PII) in datasets. ### Hugging Face — Announcing New Hugging Face and KerasHub integration - Company: Hugging Face (huggingface.co) - Announced: 2024-07-10 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/keras-hub-integration - Record: https://forck.live/items/1856-announcing-new-hugging-face-and-kerashub-integration - Subject: Platform Hugging Face and KerasHub announce a shared model save format that allows Transformers models on the Hub to be loaded directly into KerasHub, initially supporting Gemma, Llama 3, and PaliGemma models, enabling use with TensorFlow, JAX, or PyTorch backends. ### LG AI Research — [Nature Index Korea 2024] New generative AI tools to boost biomedical research - Company: LG AI Research (lgresearch.ai) - Announced: 2024-07-09 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=454 - Record: https://forck.live/items/11023-nature-index-korea-2024-new-generative-ai-tools-to-boost-biomedical-research - Subject: EXAONE LG AI Research's EXAONE is featured in Nature Index Korea 2024, highlighting its potential to assist in discovering new materials and treating diseases such as cancer and Alzheimer's. ### LG AI Research — [Nature Index Korea 2024] New generative AI tools to boost biomedical research - Company: LG AI Research (lgresearch.ai) - Announced: 2024-07-09 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=454 - Record: https://forck.live/items/4842-nature-index-korea-2024-new-generative-ai-tools-to-boost-biomedical-research - Subject: EXAONE LG AI Research announces that its EXAONE AI is featured in Nature Index Korea 2024, highlighting its potential in biomedical research. ### Hugging Face — Banque des Territoires (CDC Group) x Polyconseil x Hugging Face: Enhancing a Major French Environmental Program with a Sovereign Data Solution - Company: Hugging Face (huggingface.co) - Announced: 2024-07-09 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/sovereign-data-solution-case-study - Record: https://forck.live/items/1860-banque-des-territoires-cdc-group-x-polyconseil-x-hugging-face-enhancing-a - Subject: Platform A collaboration between Banque des Territoires (CDC Group), Polyconseil, and Hugging Face to develop a sovereign RAG-based solution to support the EduRénov environmental renovation program for French public schools. ### Hugging Face — Google Cloud TPUs made available to Hugging Face users - Company: Hugging Face (huggingface.co) - Announced: 2024-07-09 - Category: availability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/tpu-inference-endpoints-spaces - Record: https://forck.live/items/1859-google-cloud-tpus-made-available-to-hugging-face-users - Subject: Platform Hugging Face announces that Google Cloud TPUs are now available on Inference Endpoints and Spaces, with three TPU v5e instance configurations and pricing per hour. Supported models include Gemma, Llama, and Mistral. ### Hugging Face — Announcing New Dataset Search Features - Company: Hugging Face (huggingface.co) - Announced: 2024-07-08 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/datasets-filters - Record: https://forck.live/items/1861-announcing-new-dataset-search-features - Subject: Platform Hugging Face announces four new dataset search features on the Hub: filtering by modality, size, format, and library, allowing users to better discover and explore datasets.