# forck.live — 19–25 August 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-W34, 19–25 August 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-W34 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Sarvam AI — Giving voice to India's linguistic diversity with Bulbul - Company: Sarvam AI (sarvam.ai) - Announced: 2024-08-23T11:50:00+00:00 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://www.sarvam.ai/blogs/blog-bulbul/ - Record: https://forck.live/items/18406-giving-voice-to-india-s-linguistic-diversity-with-bulbul - Subject: Sarvam / Bulbul / Saaras Sarvam AI launched Bulbul v1, a text-to-speech model supporting code-mixed and multilingual speech across 10+ Indian languages. The model offers six distinct voices with consistent tone across languages and is designed for low-latency, real-time prosody control in customer-facing applications. ### LG AI Research — EXAONEPath : a Multi-Instance Learning (MIL) Framework and Patch-by-Patch Foundation Model for Processing Gigapixel-Scale Histopathology Images - Company: LG AI Research (lgresearch.ai) - Announced: 2024-08-23 - Category: open-weight-release - Coverage: not counted - Announcement: yes - Group: models - Source: https://www.lgresearch.ai/blog/view?seq=470 - Record: https://forck.live/items/4835-exaonepath-a-multi-instance-learning-mil-framework-and-patch-by-patch - Subject: EXAONE LG AI Research releases EXAONEPath, an open-source pre-trained model specialized for histopathological images, which is 1/10th the size of competing models and achieves comparable accuracy. ### Replicate — Replicate Intelligence #12 - Company: Replicate (replicate.com) - Announced: 2024-08-23 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replicate.com/blog/replicate-intelligence-2024-08-23 - Record: https://forck.live/items/2516-replicate-intelligence-12 - Subject: Platform Replicate's weekly bulletin highlighting open-source AI models, tools, and research, including FLUX.1 image generation tools, a deepfake tool, and a research paper on brain self-play during sleep. ### Hugging Face — The 5 Most Under-Rated Tools on Hugging Face - Company: Hugging Face (huggingface.co) - Announced: 2024-08-22 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/unsung-heroes - Record: https://forck.live/items/1835-the-5-most-under-rated-tools-on-hugging-face - Subject: Platform Five underrated tools on the Hugging Face Hub (ZeroGPU, Multi-process Docker, Gradio API, Webhooks, Nomic Atlas) and demonstrating how to use them to build a semantic search application. ### Vercel — Devolver ships game websites 73% faster with Vercel - Company: Vercel (vercel.com) - Announced: 2024-08-21T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/devolver-ships-game-websites-73-faster-with-vercel - Record: https://forck.live/items/20956-devolver-ships-game-websites-73-faster-with-vercel - Subject: v0 / AI SDK / AI Gateway Vercel's case study documents how Devolver, an independent game publisher, reduced system management time by over half and shipped game websites 73% faster after migrating to Vercel. The team launched five websites during a 30-minute press conference without issues. ### Supabase — Mozilla Llamafile in Supabase Edge Functions - Company: Supabase (supabase.com) - Announced: 2024-08-21 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://supabase.com/blog/mozilla-llamafile-in-supabase-edge-functions - Record: https://forck.live/items/21403-mozilla-llamafile-in-supabase-edge-functions - Subject: Supabase Supabase added Mozilla Llamafile as an inference server option for Edge Functions, alongside the existing Ollama support. Llamafile provides an OpenAI API compatible server that can be called from Supabase Edge Functions via @supabase/functions-js or the OpenAI Deno SDK. Access to Supabase-hosted open-source LLMs remains invite-only while GPU demand is managed. ### Aleph Alpha — Aleph Alpha Research and TU Darmstadt join forces to advance explainable AI in dedicated new Lab 1141 - Company: Aleph Alpha (aleph-alpha.com) - Announced: 2024-08-21 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://aleph-alpha.com/en/news/tu-darmstadt-lab-1141 - Record: https://forck.live/items/16864-aleph-alpha-research-and-tu-darmstadt-join-forces-to-advance-explainable-ai-in - Subject: Kolibri / PhariaAI Aleph Alpha Research and TU Darmstadt have formed Lab 1141, a partnership to advance explainable, safe, and transparent generative AI. The collaboration funds new PhD positions through Aleph Alpha Research Fellowships and includes Industrial PhD roles, symposiums, and internships. It aims to bridge academic research and industrial development, focusing on European values of sovereignty, explainability, and transparency. ### LG AI Research — LG Aimers Homecoming Day: Networking Party for Young AI Talents - Company: LG AI Research (lgresearch.ai) - Announced: 2024-08-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=468 - Record: https://forck.live/items/11020-lg-aimers-homecoming-day-networking-party-for-young-ai-talents - Subject: EXAONE LG AI Research held the LG Aimers Homecoming Day on August 20, bringing together over 100 participants from past hackathons for networking, mentorship from senior researchers, and meetings with HR teams from LG affiliates. The event included sessions on AI research, career advice, and business etiquette training. ### LG AI Research — LG Aimers Homecoming Day: Networking Party for Young AI Talents - Company: LG AI Research (lgresearch.ai) - Announced: 2024-08-21 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=468 - Record: https://forck.live/items/4836-lg-aimers-homecoming-day-networking-party-for-young-ai-talents - Subject: EXAONE LG AI Research hosted the LG Aimers Homecoming Day, a networking event for over 100 participants from previous hackathons, featuring mentoring from senior researchers, meetings with HR teams, and social activities. ### Hugging Face — Improving Hugging Face Training Efficiency Through Packing with Flash Attention 2 - Company: Hugging Face (huggingface.co) - Announced: 2024-08-21 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/packing-with-FA2 - Record: https://forck.live/items/1836-improving-hugging-face-training-efficiency-through-packing-with-flash - Subject: Platform Hugging Face announces a new feature in Transformers and TRL libraries that allows packing of training examples without padding, compatible with Flash Attention 2, using a new DataCollatorWithFlattening and a padding_free flag. It reports up to 2x throughput improvement and up to 20% peak memory reduction on certain datasets, with no impact on convergence. ### Harvey — Introducing the Next Version of Assistant - Company: Harvey (harvey.ai) - Announced: 2024-08-20T21:17:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.harvey.ai/blog/introducing-the-next-version-of-assistant - Record: https://forck.live/items/17731-introducing-the-next-version-of-assistant - Subject: Harvey Harvey released a revamped version of its Assistant product, adding specialized modes (Assist and Draft), iterative functionality for follow-up questions and revisions, and performance improvements including a 60% reduction in hallucinations, 23% better source accuracy, and 80% faster time to first word. The update is rolling out to all users over the next few weeks. ### Replit — Rebranding Replit: Inspiration to Action - Company: Replit (replit.com) - Announced: 2024-08-20T20:53:16+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replit.com/blog/rebranding-replit-inspiration-to-action - Record: https://forck.live/items/17992-rebranding-replit-inspiration-to-action - Subject: Replit Agent Replit announced a rebranding of its visual identity, drawing inspiration from the history of software development and the role of AI in democratizing software creation. The new brand emphasizes the connection between code and language, and positions AI as a continuation of the historical trend of making software more accessible. ### OpenAI — OpenAI partners with Condé Nast - Company: OpenAI (openai.com) - Announced: 2024-08-20T11:00:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/conde-nast - Record: https://forck.live/items/719-openai-partners-with-cond-nast - Subject: GPT / ChatGPT / API ### OpenAI — Putting AI to work at Upwork - Company: OpenAI (openai.com) - Announced: 2024-08-20T10:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/upwork - Record: https://forck.live/items/720-putting-ai-to-work-at-upwork - Subject: GPT / ChatGPT / API Upwork is using AI to unite team members, operations, and product development. ### Hugging Face — Deploy Meta Llama 3.1 405B on Google Cloud Vertex AI - Company: Hugging Face (huggingface.co) - Announced: 2024-08-19 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/llama31-on-vertex-ai - Record: https://forck.live/items/1837-deploy-meta-llama-3-1-405b-on-google-cloud-vertex-ai - Subject: Platform Deploying the Meta Llama 3.1 405B model (FP8 quantized variant) on Google Cloud Vertex AI using Hugging Face Deep Learning Containers and Text Generation Inference.