A video is trending on social media that makes some thing of “impersonation attacks”. My take: just „clickbait“, and not a real threat: inter-frame variance too high, caused by different hallucination artifacts introduced into each individual frame. See screenshots. very low-frame rate, obviously. Scaling it up may introduce even more artif... Read more 19 Oct 2023 - 1 minute read
Summary: The team is positioned early in the market; no blind spots identified at this stage. Key Points: Discussion with a VIP customer: Awareness of AI’s role in fraud detection and cyber threat combat. Interest was piqued when presented with our ambitions, especially with our International Markets flagship product. ... Read more 17 Oct 2023 - 1 minute read
There is some concern over whether companies and content creators are, or should be, in agreement with their data being utilized for training AI models. The Google C4 dataset, a massive collection of web-crawled data, is at the center of this debate. However, a very recent trend is to move away from web resources like C4. Instead, high-quality,... Read more 12 Oct 2023 - 1 minute read
Upfront summary: model training has evolved significantly. Adding new knowledge may actually be viable now. Background: a recent article shared on LinkedIn about Google Cloud Next had caught my attention. Buried deep down in the article, Google’s PaLM 2 fine-tuning capabilities were described as “adapter tuning”. 💡 Adapter-based Tuning of LLMs ... Read more 06 Oct 2023 - 5 minute read
Amazon Bedrock is Generally Available: Announcement. For everyone who has enjoyed preview access, these details caught my attention: “Amazon Titan model usage will be billed starting 00:00 PST on Sept 28, 2023.” “Effective Sept. 18, 2023: Titan Text G1 - Express is the new name for model Titan Text Large.” Model overview: The only model... Read more 29 Sep 2023 - less than 1 minute read