Thousand Questions on Tongyi

On March 11th, Manus official microblogging announced that it has officially reached a strategic cooperation with Ali Tongyi Qianqian team. Based on the open-source models of Tongyi Qianqian series, the two parties will realize all the functions of Manus on the domestic models and computing power platform. The two technical teams have been working closely together to create more creative general intelligence products for Chinese users. It is understood that in November 2024, the company completed its Series A financing, led by Tencent and Sequoia China, followed by Zhenge Fund and Wang Huiwen. Previously, the company had in February 2023 and 8 ...

March 6, 2011 - Research has shown that reinforcement learning can significantly improve the inference capabilities of models such as DeepSeek-R1, which achieves state-of-the-art performance by integrating cold-start data and multi-stage training, enabling deep thinking and complex inference. AliCloud Tongyi Qianqi officials today announced the availability of their latest inference model, QwQ-32B, a model with 32 billion parameters that rivals the performance of DeepSeek-R1 with 671 billion parameters, 37 billion of which are activated. This achievement highlights the importance of bringing reinforcement learning to ...

January 29, 2011 - On the occasion of the New Year, AliCloud announced its new Qwen 2.5-Max hyperscale MoE model, which can be accessed by way of APIs, or you can log in to Qwen Chat to experience it, for example, by talking to the model directly, or using artifacts, search and other functions. According to the introduction, Qwen 2.5-Max uses over 20 trillion tokens of pre-training data and a well-designed post-training program for training. Performance AliCloud directly compares ...

January 7 news, 1AI from the official WeChat public number of Ali Cloud was informed that Transn Holdings (hereinafter referred to as Transn) has reached a cooperation with Ali Cloud, Tongyi Qianqian big model has been carried in the AI cell phone PHANTOM V Fold2 launched by TECNO, the technology brand of Transn, to create a deep localization of the "practical AI". "AI Through the specific AI buttons on the body, users can call the end-side large model with one key, and can still run multiple rounds of AI conversations smoothly in an offline environment, and realize document and call summaries. According to reports, Transn and Aliyun on the phone...

On January 4, Ali Tongyi Qianqian Qwen newly launched the CodeElo benchmark test to assess the programming level of Large Language Model (LLM) by comparing the Elo rating system with human programmers. Project Background One of the AI scenario applications of Large Language Model is to generate and complement code, but there are many challenges in assessing the real ability of programming at this stage. Existing benchmarks including LiveCodeBench and USACO have limitations, lack robust private test cases, do not support specialized judgment systems,...

Ali Tongyi Qwen team released a blog post today (December 25th), announcing the launch of QVQ-72B-Preview, an open source visual reasoning model based on the Qwen2-VL-72B build, which is capable of finding solutions through logical reasoning calmly in the face of complex physics problems like a master of physics. Ali Tongyi Thousand Questions team evaluates QVQ-72B-Preview on 4 datasets, and 1AI attaches the relevant introduction as follows: MMMU: A university-level multidisciplinary and multimodal evaluation set designed to examine the model visual...

The former Ali Tongyi Qianqian big model technology leader Zhou Chang (fancy name: Zhong Huang) was exposed on July 18th will leave to start a business, but on October 23rd, there is news that Zhou Chang has low-profile to join the byte jump. On November 13, a news about Alibaba's decision to apply for arbitration "Tongyi big model former employee Zhou Chang violated the non-compete agreement" quickly spread in the AI circle, close to Tongyi's industry insiders said "the situation is true". According to the first financial report today, a service byte headhunting company staff revealed: "Resignation to start a business is a front, is afraid of being non-compete, but...

According to surging news reports, Alibaba Group CEO Wu Yongming delivered a speech at the plenary session, saying that AI promotes productivity changes in thousands of industries and will create value far beyond that created by Internet links, "facing the AI era, Alibaba focuses on doing two things well, one is to continuously improve AI infrastructure, and the other is to adhere to open source and openness, and transform AI capabilities into productivity in thousands of industries. Productivity." Wu Yongming further revealed that in the past year, Alibaba has made intense investment in AI infrastructure. Thanks to the scale effect and technological progress, the arithmetic cost continues to decrease. Among them, Alibaba ...

November 19th, Ali Tongyi Qianqian released a blog post yesterday (November 18th) announcing the launch of the Qwen2.5-Turbo open source AI model in response to the community's request for a longer Context Length after months of optimization and polishing. Qwen2.5-Turbo extends the context length from 128,000 to 1,000,000 tokens, an improvement equivalent to about 1,000,000 English words or 1,500,000 Chinese characters, and can accommodate 10 complete novels,...

November 12 news, Ali Tongyi Thousand Questions open source Qwen2.5-Coder full series of models, of which Qwen2.5-Coder-32B-Instruct become the current SOTA open source model, the official claim that the code ability to level with the GPT-4o. Qwen2.5-Coder-32B-Instruct as the open-source flagship model, on several popular code generation benchmarks (such as EvalPlus, LiveCodeBench, BigCodeBench) are...

At the 2024 Cloud Root Conference, AliCloud CTO Zhou Jingren released Qwen2.5, a new generation of open source models, of which the flagship model, Qwen2.5-72B, is claimed to outperform the Llama 405B. Qwen2.5 covers multiple sizes of large language models, multimodal models, mathematical models, and code models, and each size has a base version, a command-following version, quantized versions, totaling more than 100 models on the shelf. Qwen2.5 language models: 0.5B, 1.5B, 3B, 7B, 14B, 32B, and ...

Alibaba's "Tongyi Qianwen" team has made another big news! They just released Qwen2Math Demo, this math model is a little monster, even GPT-4 is trampled under its feet. This model can not only handle math problems input by text, but also understand formulas in pictures and screenshots. Imagine that you take a photo of a formula and it can give you the answer. It is simply a magic tool for math class! (Of course, we do not advocate cheating) Qwen2-Math has launched three versions: 72B, 7B and 1.5B. Among them, 7…

Alibaba's large language model "Tongyi Qianwen" announced the launch of a new domain name "tongyi.ai" and brought a number of new features. Summarized as follows: The web version of the chat adds a new deep search function: supports more content source indexes, search results are more in-depth, professional and structured, and the digital corner mark floats to display the source web page. App image micro-motion effects support multi-size images: enter the Tongyi App channel page, select "Image Micro-motion Effects", you need to upload a picture, and you can generate sound effects and micro-motion video effects that match the picture. App custom singing and acting supports 3:4 aspect ratio (originally 1:1): audio...

Ali Tongyi Qianwen has open-sourced two models of the Qwen2-Audio series, Qwen2-Audio-7B and Qwen2-Audio-7B-Instruct. As a large-scale audio language model, Qwen2-Audio is able to accept various audio signal inputs and perform audio analysis or directly respond to text based on voice instructions. It has two different audio interaction modes: Voice chat: users can freely interact with Qwen2-Audio by voice without text input Audio analysis: users can…

Ali Tongyi Qianwen Qwen2 open source family welcomes a new member Qwen2-Math, which has three versions with 1.5 billion parameters, 7 billion parameters and 72 billion parameters. It is a language model built on Qwen2 LLM and specially used for solving mathematical problems. Introduction Qwen2-Math is a series of language models built on Qwen2 LLM and specially used for solving mathematical problems. Its mathematical ability significantly exceeds that of open source models and even closed source models (such as GPT-4o). The official hopes to help the scientific community solve problems that require complex multi-step logical reasoning…

The Paris Olympics is about to open. As an AI technology provider, Alibaba Group's Tongyi Big Model will bring revolutionary changes to the event. At the 142nd General Assembly of the International Olympic Committee, Alibaba and technology giants such as Intel were confirmed as partners of the Olympic AI Agenda. Alibaba's AI technology will enhance the Olympic experience in many aspects. The 360-degree live broadcast special effects will provide viewers with a full range of immersive viewing experience. At the same time, the black and white image AI color restoration technology will give historical images a new lease of life and present a more vivid visual feast for the audience. In addition, the application of carbon emission reduction auxiliary technology will help achieve green...

Alibaba Cloud Tongyi Qianwen has open-sourced two voice base models: SenseVoice (for voice recognition) and CosyVoice (for voice generation). SenseVoice focuses on high-precision multi-language voice recognition, emotion recognition, and audio event detection, and has the following features: Multi-language recognition: uses more than 400,000 hours of data training, supports more than 50 languages, and has better recognition results than the Whisper model Rich text recognition: has excellent emotion recognition, and can achieve and exceed the results of the current best emotion recognition model on test data; supports...

Demand scenarios When you want to publish a video with a unified cover image, you often can't find a cover image with a unified style for a while. Although it is okay to temporarily pick a picture you like, the cover effect of multiple videos appears very messy, giving people a feeling of lack of care and professionalism. Solution It is planned to generate 20 common fruit pictures in advance, with a screen ratio of 16:9 and 9:16, which will be used for the horizontal screen cover of Xigua Video and the vertical screen cover of Douyin Video respectively. Use the tools Tongyi Qianwen + Tongyi Wanxiang. Prompt word example: The main body of the picture is an apple, front view, which can be seen clearly and completely...

When I was a student, I read "Dream of Red Mansions". The characters in the novel have always existed in my mind, but they are rather vague. After studying AI for a while, I decided to visualize the characters in the classic novels to see what they look like in the eyes of AI? How about the finished film drawn by AI? The AI tool used this time: Alibaba's Tongyi Qianwen. https://www.1ai.net/6763.html Let's take a look at the effect first: Now, let's learn how to use Tongyi to generate photos of classic novel characters! How to use Tongyi to generate photos of classic novel characters? Before generating photos of novel characters, first confirm it...

Qwen announced today that after months of hard work, the Qwen series model has been significantly upgraded from Qwen1.5 to Qwen2, and has been simultaneously open sourced on Hugging Face and ModelScope. The main contents of Qwen 2.0 are as follows: 5 sizes of pre-trained and instruction fine-tuned models, including Qwen2-0.5B, Qwen2-1.5B, Qwen2-7B, Qwen2-57B-A14B and Qwen2-72B Based on Chinese and English…

ByteDance officially released the Doubao large model on May 15, announcing that the price of the large model has entered the "centimeter era", claiming that it is "99.3% cheaper than the industry". Today, Alibaba Cloud followed closely and dropped a bombshell: Tongyi Qianwen GPT-4-level main model Qwen-Long announced a price cut, and the API input price dropped from 0.02 yuan/thousand tokens to 0.0005 yuan/thousand tokens, a drop of 97%. In short, 1 yuan can buy 2 million tokens, which is equivalent to 5 copies of Xinhua...

At the Alibaba Cloud AI Leaders Summit-Beijing, the Tongyi Qianwen 2.5 large model was released, claiming to surpass GPT-4 in many capabilities. According to Alibaba Cloud officials, the Tongyi large model has served more than 90,000 enterprises through Alibaba Cloud, and the cumulative downloads of Tongyi open source models have exceeded 7 million. Tongyi's application process has accelerated and has now entered the fields of PC, mobile phones, and automobiles. At the event, Alibaba Cloud officially released Tongyi Qianwen 2.5, claiming to "upgrade capabilities and surpass GPT-4 in all aspects". In the Chinese context, text understanding, text generation, knowledge question answering & ...

At the Alibaba Cloud AI Leaders Summit today, Alibaba Cloud Chief Technology Officer (CTO) Zhou Jingren revealed a remarkable statistic: Tongyi Qianwen's daily API call volume has exceeded 100 million times, the number of corporate users has successfully surpassed 90,000, and the number of open source model downloads has reached an astonishing 7 million times. Zhou Jingren further introduced Tongyi Qianwen's professional performance in the industry. He pointed out that after strict evaluation by a third-party organization, Tongyi Qianwen has gradually approached and approached GPT4 in terms of performance, becoming the industry leader. Especially in terms of knowledge questions and answers in the Chinese context, Tongyi Qianwen has shown...

Image from ideogram generation Today (March 16) is the time for Guangxi's 2024 civil servant written examination. I heard some friends discussing the positions they applied for. Guotou thought: There are so many domestic AI models now. Is it possible to use them to help me quickly interpret the information in the job list? Which one is better, Wenxin Yiyan 3.5, iFlytek Spark, Doubao, Zhipu Qingyan GLM-3, Zhipu Qingyan GLM-4, Tongyi Qianwen, or Kimi? I went through my own needs in my mind, as follows: Upload the job table excel file. Verify whether AI can read the table data correctly...