link1s.site

Stanford AI project team apologizes for plagiarizing Chinese model

We apologize to the authors of MiniCPM [the AI model developed by a Chinese company] for any inconvenience that we caused for not doing the full diligence to verify and peer review the novelty of this work, the multimodal AI model Llama3-V's developers wrote in a post on social platform X.

The apology came after the team from Stanford University announced Llama3-V on May 29, claiming it had comparable performance to GPT4-V and other models with the capability to train for less than $500.

According to media reports, the announcement published by one of the team members quickly received more than 300,000 views.

However, some netizens from X found and listed evidence of how the Llama3-V project code was reformatted and similar to MiniCPM-Llama3-V 2.5, an LLM developed by a Chinese technology company, ModelBest, and Tsinghua University.

Two team members, Aksh Garg and Siddharth Sharma, reposted a netizen's query and apologized on Monday, while claiming that their role was to promote the model on Medium and X (formerly Twitter), and that they had been unable to contact the member who wrote the code for the project.

They looked at recent papers to validate the novelty of the work but had not been informed of or were aware of any of the work by Open Lab for Big Model Base, which was founded by the Natural Language Processing Lab at Tsinghua University and ModelBest, according to their responses. They noted that they have taken all references to Llama3-V down in respect to the original work.

In response, Liu Zhiyuan, chief scientist at ModelBest, spoke out on the Chinese social media platform Zhihu, saying that the Llama3-V team failed to comply with open-source protocols for respecting and honoring the achievements of previous researchers, thus seriously undermining the cornerstone of open-source sharing.

According to a screenshot leaked online, Li Dahai, CEO of ModelBest, also made a post on his WeChat moment, saying that the two models were verified to have highly similarity in terms of providing answers and even the same errors, and that some relevant data had not yet been released to the public.

He said the team hopes that their work will receive more attention and recognition, but not in this way. He also called for an open, cooperative and trusting community environment.

Director of the Stanford Artificial Intelligence Laboratory Christopher Manning also responded to Garg's explanation on Sunday, commenting "How not to own your mistakes!" on X.

As the incident became a trending topic on Sina Weibo, Chinese netizens commented that academic research should be factual, but the incident also proves that the technology development in China is progressing.

Global Times

TikTok to introduce a new feature that can clone your voice with AI in just 10 second

Use of AI is certainly the hottest topic in the tech industry and every major and minor player in this industry is using AI in some way. Tools like ChatGPT can help you do a wide range of task and even help you generate images. The other thing is - Voice Cloning. OpenAI recently introduced a voice engine that can generate clone of your voice with just 15 seconds of your audio. There is no shortage of voice cloning tools on the web which can help you do the same. The newest tech giant which is going to use AI to clone your voice is - TikTok. We all know TikTok, posting short videos with filters, effects and all other kind of things. So TikTok found a way to use the voice cloning AI in its app. TikTok is working on this feature, which does not seem to really have a proper name, it just references it as "Create your voice with AI" and "TikTok Voice Library". In the latest version of TikTok I came across some strings which indicates that TikTok is working on it. I was also able to access the initial UI which introduces the feature and was able to see the terms and condition of "TikTok Voice Library" which user have to accept in order to use the feature. Here are the screenshots from the app- As you can in the screenshot above, this is the initial screen which a user will see for the first time they access this feature. Tiktok claims that it can create an AI verison of your voice in just 10 seconds. The generated AI voice clone can be used with text-to-speech in TikTok videos. It also outline the process of how it will work. You have to record yourself speaking and TikTok will process the voice and use information about your voice to generate your AI voice. When it comes to privacy, your AI voice will stay private and you can delete it anytime. Tapping the "Continue" button brings "TikTok Voice Library Terms" screen which a user should definitely read, you can see here and read as well - How it will work After agreeing to terms and conditions I was introduced with a screen where TikTok will show some text and user have to press the record button while reading the text. Now unfortunately I did not see any text. This is probably because the feature is not fully ready or the backend from which it fetches the text is not live yet. Manually pressing the record button and saying random things also shows an error. So, it's also not possible to provide any sample voice generated with it and see how it compares to other voice cloning competitors. If it starts working someday, it will process your recorded voice and generate AI version of your voice. Here is a screenshot of that screen - My guess is that whenever the feature starts working, users have to clone voice only one time and the saved AI voice can be used through the text-to-speech method to add voice in your videos. You just have to type the words, choice is yours :p

Stanford AI project team apologizes for plagiarizing Chinese model

An artificial intelligence (AI) team at Stanford University apologized for plagiarizing a large language model (LLM) from a Chinese AI company, which became a trending topic on the Chinese social media platforms, where it sparked concern among netizens on Tuesday. We apologize to the authors of MiniCPM [the AI model developed by a Chinese company] for any inconvenience that we caused for not doing the full diligence to verify and peer review the novelty of this work, the multimodal AI model Llama3-V's developers wrote in a post on social platform X. The apology came after the team from Stanford University announced Llama3-V on May 29, claiming it had comparable performance to GPT4-V and other models with the capability to train for less than $500. According to media reports, the announcement published by one of the team members quickly received more than 300,000 views. However, some netizens from X found and listed evidence of how the Llama3-V project code was reformatted and similar to MiniCPM-Llama3-V 2.5, an LLM developed by a Chinese technology company, ModelBest, and Tsinghua University. Two team members, Aksh Garg and Siddharth Sharma, reposted a netizen's query and apologized on Monday, while claiming that their role was to promote the model on Medium and X (formerly Twitter), and that they had been unable to contact the member who wrote the code for the project. They looked at recent papers to validate the novelty of the work but had not been informed of or were aware of any of the work by Open Lab for Big Model Base, which was founded by the Natural Language Processing Lab at Tsinghua University and ModelBest, according to their responses. They noted that they have taken all references to Llama3-V down in respect to the original work. In response, Liu Zhiyuan, chief scientist at ModelBest, spoke out on the Chinese social media platform Zhihu, saying that the Llama3-V team failed to comply with open-source protocols for respecting and honoring the achievements of previous researchers, thus seriously undermining the cornerstone of open-source sharing. According to a screenshot leaked online, Li Dahai, CEO of ModelBest, also made a post on his WeChat moment, saying that the two models were verified to have highly similarity in terms of providing answers and even the same errors, and that some relevant data had not yet been released to the public. He said the team hopes that their work will receive more attention and recognition, but not in this way. He also called for an open, cooperative and trusting community environment. Director of the Stanford Artificial Intelligence Laboratory Christopher Manning also responded to Garg's explanation on Sunday, commenting "How not to own your mistakes!" on X. As the incident became a trending topic on Sina Weibo, Chinese netizens commented that academic research should be factual, but the incident also proves that the technology development in China is progressing. Global Times

NHTSA opens recall query into about 94,000 Jeep Wrangler 4xe SUVs

July 9 (Reuters) - The National Highway Traffic Safety Administration (NHTSA) has opened a recall query into 94,275 Stellantis-owned (STLAM.MI), opens new tab Jeep SUVs over a loss of motive power, the U.S. auto safety regulator said on Tuesday. The investigation targets Jeep's Wrangler 4xe hybrid SUVs manufactured between 2021 through 2024. Chrysler had previously recalled, opens new tab the same model in 2022 to address concerns related to an engine shutdown. A recall query is an investigation opened by safety regulators when a remedy to solve an issue appears inadequate. The complaints noted in the new report include both failures in vehicles that received the recall remedy and those not covered by the prior recall, the NHTSA said.

Xinjiang scientists discover plant with potential to survive on Mars

In a groundbreaking discovery, researchers from the Xinjiang Institute of Ecology and Geography of the Chinese Academy of Sciences have found a desert moss species, known as Syntrichia caninervis, that has the potential to survive in the extreme conditions on Mars. The Global Times learned from the institute that during the third Xinjiang scientific expedition, the research team focused on studying the desert moss and found that it not only challenges people's understanding of the tolerance of organisms in extreme environments, but also demonstrates the ability to survive and regenerate under simulated Martian conditions. Supported by the Xinjiang scientific expedition project, researchers Li Xiaoshuang, Zhang Daoyuan and Zhang Yuanming from the Xinjiang Institute of Ecology and Geography and Kuang Tingyun, an academician from the Chinese Academy of Sciences, concentrated on studying the "pioneer species" Syntrichia caninervis in an extreme desert environment, according to the institute in an article it sent to the Global Times on Sunday. Through scientific experiments, the researchers systematically proved that the moss can tolerate over 98 percent cell dehydration, survive at temperatures as low as -196 C without dying, withstand over 5000Gy of gamma radiation without perishing, and quickly recover, turn green, and resume growth, showcasing extraordinary resilience. These findings push the boundaries of human knowledge on the tolerance of organisms in extreme environments. Furthermore, the research revealed that under simulated Martian conditions with multiple adversities, Syntrichia caninervis can still survive and regenerate when returned to suitable conditions. This marks the first report of higher plants surviving under simulated Martian conditions. The research team also identified unique characteristics of Syntrichia caninervis. Its overlapping leaves reduce water evaporation, while the white tips of the leaves reflect intense sunlight. Additionally, the innovative "top-down" water absorption mode of the white tips efficiently collects and transports water from the atmosphere. Moreover, the moss can enter a selective metabolic dormancy state in adverse environments and rapidly provide the energy needed for recovery when its surrounding environment improves. Based on the extreme environmental tolerance of Syntrichia caninervis, the research team plans to conduct experiments on spacecraft to monitor the survival response and adaptation capabilities of the species under microgravity and various ionizing radiation adversities. They aim to unravel the physiological and molecular basis of the moss and explore the key life tolerance regulatory mechanisms, laying the foundation for future applications of Syntrichia caninervis in outer space colonization.

Iran's president-elect reaffirms policy toward Israel

Iran's President-elect Masoud Pezeshkian reiterated Iran's anti-Israel stance on Monday, saying resistance movements across the region will not allow Israel's "criminal policies" against Palestinians to continue. Pezeshkian told Hassan Nasrallah, the leader of Iran-backed Lebanese Hezbollah, that "the Islamic Republic will always support the people of the region in their resistance against the illegal Zionist regime." This suggests that the incoming government will not change its regional policy under the relatively moderate Pezeshkian, who defeated his hard-line opponent in a runoff election last week. Pezeshkian was quoted as saying by Iranian media, "I am sure that the regional resistance movement will not allow this regime to continue its militant and criminal policies against the oppressed people of Palestine and other countries in the region." The Shiite Muslim Hezbollah and the Palestinian Sunni Muslim Hamas are both part of the local "resistance axis" faction organization supported by Iran. Israel did not immediately comment on Pezeshkian's speech. Hamas led an attack on southern Israel on October 7. According to Israeli statistics, Hamas killed 1,200 people and kidnapped about 250 hostages, triggering the Israeli-Palestinian war. The Gaza Health Ministry said that the Israeli military attack killed more than 38,000 Palestinians and injured nearly 88,000 people.