6/ IT3D – presents a strategy that leverages explicitly synthesized multi-view images to improve Text-to-3D generation; integrates a discriminator along a Diffusion-GAN dual training strategy to guide the training of the 3D models.
AI
-
Survey on Instruction Tuning Methods for Large Language Models
By
–
2/ Survey on Instruction Tuning for LLMs – new survey paper on instruction tuning LLM, including a systematic review of the literature, methodologies, dataset construction, training models, applications, and more.
-
SeamlessM4T: Unified Multilingual Multimodal Machine Translation System
By
–
3/ SeamlessM4T – a unified multilingual and multimodal machine translation system that supports ASR, text-to-text translation, speech-to-text translation, text-to-speech translation, and speech-to-speech translation.
-
LLM Security: Identifying Threats and Building Robust Systems
By
–
4/ Use of LLMs for Illicit Purposes – provides an overview of existing efforts to identify and mitigate threats and vulnerabilities arising from LLMs; serves as a guide to building more reliable and robust LLM-powered systems.
-
LLMs Water Consumption Raises Ecocide Concerns Globally
By
–
Growing number of countries consider making ecocide a crime https://
theguardian.com/environment/20
23/aug/26/growing-number-of-countries-consider-making-ecocide-crime
… Where its well documented LLMs use massive amounts of water drastically harming people and planet, let’s call the knowledge of their harm what it is: justified ecocide. Nope. -
Chief AI Officer: Strategic Leadership Role in Modern Business
By
–
Unveiling the pivotal role of a #Chief #AI #Officer, this article delves into why #businesses must embed AI in their core #strategy and the #skills required for this #transformative #leadership position.
-
FAccT Conference Videos: Alondra’s Thick Alignment Provocation
By
–
For all of us needing to catch up on the work @FAccTConference this year, all the videos are online now I’m starting by catching up on @alondra
’s provocation on thick alignment -
Weights compatibility and numerical stability in LoRA fine-tuning
By
–
Yes, I think that's either maybe so that (1) people with older cards can use these weights as well and (2) we maybe have more stability when finetuning with LoRA etc (in case numbers become large)
It's interesting for sure -
Four Levels of Cyber Resilience from Champions to Blockers
By
–
The four levels of cyber resilience range from Business Blockers, prioritizing security over strategy, to Vulnerable firms with minimal protection. Cyber Champions balance resilience with goals, while Risk Lovers prioritize growth, accepting higher risk. Microblog @antgrasso
-
Military AI, Henry Kissinger, and the Age of AI Hype
By
–
And the winner (first across 1k views) is … Episode 14! “Henry Kissinger, Machines of War, and the Age of Military AI Hype” https://
peertube.dair-institute.org/w/h33StnpVknj2
wJhrthS2kS
…