1.
zotero 插件
1.
zotero-better-notes 插件 for zotero
2.
3.
4.
dark mode change
2.
3.
4.
Google DeepMind researchers have discovered 2.2mn crystal structures that open potential progress in fields from renewable energy to advanced computation, and show the power of artificial intelligence to discover novel materials. The trove of theoretically stable but experimentally unrealised combinations identified using an AI tool known as GNoME is more than 45 times larger than the number of such substances unearthed in the history of science, according to a paper published in Nature on Wednesday.
About 20,000 of the crystals experimentally identified in the ICSD database are computationally stable. Computational approaches drawing from the Materials Project, Open Quantum Materials Database and WBM database boosted this number to 48,000 stable crystals. GNoME expands the number of stable materials known to humanity to 421,000.
We scale pretraining of a NequIP potential30 on data sampled from ionic relaxations. Increasing the pretraining dataset, we observe consistent power-law improvements in accuracy (see Fig. 3a,b). Despite only being trained on ionic relaxations and not on molecular-dynamics data, the pretrained GNoME potential shows remarkable accuracy when evaluated on downstream data sampled from the new distribution of AIMD in a zero-shot manner, that is, in which no training data originate from AIMD simulations (see Fig. 3). Notably, this includes unseen compositions, melted structures and structures including vacancies, all of which are not included in our training set (see Supplementary Note 6.4). In particular, we find that the scale of the GNoME dataset allows it to outperform existing general-purpose potentials (see Fig. 3d) and makes the pretrained potential competitive with models trained explicitly on hundreds of samples from the target data distributions (see Supplementary Note 6.4). We observe particularly pronounced improvements in the transferability of MLIPs, one of the most pressing shortcomings of MLIPs. To assess the transferability of the potentials, we test their performance under distribution shift: we train two types of NequIP potential on structures sampled from AIMD at T = 400 K, one in which the network is trained from randomly initialized weights and the other in which we fine-tune from a pretrained GNoME checkpoint. We then measure the performance of both potentials on data sampled from AIMD at T = 1,000 K (see Fig. 3c), out of distribution with respective to the 400-K data. The potential pretrained on GNoME data shows systematic and strong improvements in transferability over the potential trained from scratch, even when training is performed on more than 1,000 structures. The zero-shot GNoME potential, not fine-tuned on any data from this composition, outperforms even a state-of-the-art NequIP model trained on hundreds of structures.
5.
Samsung internet for galaxy windows
6.
Top Shows
- Crime Junkie
- The Daily
- Dateline NBC
- SmartLess
- This American Life
- Morbid
- Up First
- Huberman Lab
- Hidden Brain
- Stuff You Should Know
7.
para
Message from Sam to the company
I am returning to OpenAI as CEO. Mira will return to her role as CTO. The new initial board will consist of Bret Taylor (Chair), Larry Summers, and Adam D’Angelo.
I have never been more excited about the future. I am extremely grateful for everyone’s hard work in an unclear and unprecedented situation, and I believe our resilience and spirit set us apart in the industry. I feel so, so good about our probability of success for achieving our mission.
Before getting to what comes next, I’d like to share some thanks.
I love and respect Ilya, I think he's a guiding light of the field and a gem of a human being. I harbor zero ill will towards him. While Ilya will no longer serve on the board, we hope to continue our working relationship and are discussing how he can continue his work at OpenAI.
I am grateful to Adam, Tasha, and Helen for working with us to come to this solution that best serves the mission. I’m excited to continue to work with Adam and am sincerely thankful to Helen and Tasha for investing a huge amount of effort in this process.
Thank you also to Emmett who had a key and constructive role in helping us reach this outcome. Emmett’s dedication to AI safety and balancing stakeholders’ interests was clear.
Mira did an amazing job throughout all of this, serving the mission, the team, and the company selflessly throughout. She is an incredible leader and OpenAI would not be OpenAI without her. Thank you.
Greg and I are partners in running this company. We have never quite figured out how to communicate that on the org chart, but we will. In the meantime, I just wanted to make it clear. Thank you for everything you have done since the very beginning, and for how you handled things from the moment this started and over the last week.
The leadership team–Mira, Brad, Jason, Che, Hannah, Diane, Anna, Bob, Srinivas, Matt, Lilian, Miles, Jan, Wojciech, John, Jonathan, Pat, and many more–is clearly ready to run the company without me. They say one way to evaluate a CEO is how you pick and train your potential successors; on that metric I am doing far better than I realized. It’s clear to me that the company is in great hands, and I hope this is abundantly clear to everyone. Thank you all.
Jakub, Szymon, and Aleksander are exceptional talents and I’m so happy they have rejoined to move us and our research forward. Thank you.
To all of you, our team: I am sure books are going to be written about this time period, and I hope the first thing they say is how amazing the entire team has been. Now that we’re through all of this, we didn’t lose a single employee. You stood firm for each other, this company, and our mission. One of the most important things for the team that builds AGI safely is the ability to handle stressful and uncertain situations, and maintain good judgment throughout. Top marks. Thank you all.
Satya, Kevin, Amy, and Brad have been incredible partners throughout this, with exactly the right priorities all the way through. They’ve had our backs and were ready to welcome all of us if we couldn’t achieve our primary goal. We clearly made the right choice to partner with Microsoft and I’m excited that our new board will include them as a non-voting observer. Thank you.
To our partners and users, thank you for sticking with us. We really felt the outpouring of support and love, and it helped all of us get through this. The fact that we did not lose a single customer will drive us to work even harder for you, and we are all excited to get back to work.
Will Hurd, Brian Chesky, Bret Taylor and Larry Summers put their lives on hold and did an incredible amount to support the mission. I don’t know how they did it so well, but they really did. Thank you.
Ollie also put his life on hold this entire time to just do everything he could to help out, in addition to providing his usual unconditional love and support. Thank you and I love you.
So what’s next?
We have three immediate priorities.
Advancing our research plan and further investing in our full-stack safety efforts, which have always been critical to our work. Our research roadmap is clear; this was a wonderfully focusing time. I share the excitement you all feel; we will turn this crisis into an opportunity! I’ll work with Mira on this.
Continuing to improve and deploy our products and serve our customers. It’s important that people get to experience the benefits and promise of AI, and have the opportunity to shape it. We continue to believe that great products are the best way to do this. I’ll work with Brad, Jason and Anna to ensure our unwavering commitment to users, customers, partners and governments around the world is clear.
Bret, Larry, and Adam will be working very hard on the extremely important task of building out a board of diverse perspectives, improving our governance structure and overseeing an independent review of recent events. I look forward to working closely with them on these crucial steps so everyone can be confident in the stability of OpenAI.
I am so looking forward to finishing the job of building beneficial AGI with you all—best team in the world, best mission in the world.
Love,
Sam
Message from Bret to the company
On behalf of the OpenAI Board, I want to express our gratitude to the entire OpenAI community, especially all the OpenAI employees, who came together to help find a path forward for the company over the past week. Your efforts helped enable this incredible organization to continue to serve its mission to ensure that artificial general intelligence benefits all of humanity. We are thrilled that Sam, Mira and Greg are back together leading the company and driving it forward. We look forward to working with them and all of you.
As a Board, we are focused on strengthening OpenAI’s corporate governance. Here’s how we plan to do it:
- We will build a qualified, diverse Board of exceptional individuals whose collective experience represents the breadth of OpenAI’s mission – from technology to safety to policy. We are pleased that this Board will include a non-voting observer for Microsoft.
- We will further stabilize the OpenAI organization so that we can continue to serve our mission. This will include convening an independent committee of the Board to oversee a review of the recent events.
- We will enhance the governance structure of OpenAI so that all stakeholders – users, customers, employees, partners, and community members – can trust that OpenAI will continue to thrive.
OpenAI is a more important institution than ever before. ChatGPT has made artificial intelligence a part of daily life for hundreds of millions of people. Its popularity has made AI – its benefits and its risks – central to virtually every conversation about the future of governments, business, and society.
We understand the gravity of these discussions and the central role of OpenAI in the development and safety of these awe-inspiring new technologies. Each of you plays a critical part in ensuring that we effectively meet these challenges. We are committed to listening and learning from you, and I hope to speak with you all very soon.
We are grateful to be a part of OpenAI, and excited to work with all of you.
Thank you,
Bret Taylor
Chair, OpenAI
8.
9.
carinsurance company
10.
11.
12.
Deno Cron
13.
新技术用超声波芯片而不是电来发开脑机接口
BCI developers
Neuralink，Paradromics and Synchron
all reached important milestones in the clinical trial process, each bringing a fully implantable BCI a little closer to clinical reality in the United States. Those companies are developing BCIs that interact with the brain electrically, but a recently announced research collaboration is developing a BCI that uses a radically different medium: ultrasound
companies Forest Neurotech and Butterfly Network
解释 For stimulation, focused ultrasound waves can flexibly target specific regions throughout the brain from outside the skull. In contrast, electrical stimulation techniques are more spatially limited because electricity does not travel very far through brain tissue, so electrically stimulating deep regions of the brain requires invasive surgical procedures that place electrodes near the relevant sites. Ultrasound can both record and control neural activity.
14.
Anthropic数据训练集
通过 Amazon Mechanical Turk 上的聊天工具生成
15.
RLHF
利用 Reward Model 输出的奖励，用强化学习方式微调优化 Language Model
策略梯度强化学习 (Policy Gradient RL) 算法、近端策略优化 (Proximal Policy Optimization，PPO) 微调初始 LM 的部分或全部参数
因为微调整个 10B～100B+ 参数的成本过高 (相关工作参考低秩适应 LoRA 和 DeepMind 的 Sparrow LM) 。PPO 算法已经存在了相对较长的时间，有大量关于其原理的指南，因而成为 RLHF 中的有利选择
RLHF eg. project (RLHF first programme)
tl;dr
16.
Zepz
In May, Zepz cut 26% of its workforce, citing duplication of roles that resulted from its acquisition of Sendwave, another money transfer service.
The money transfer group that owns WorldRemit,made a fresh round of layoffs.
没有很多内容，捕捉到的信息很少
