Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

AlphaCode was the first AI to write code at a human level in competitive programming. 🛠 Using a specialized version of Gemini, we created AlphaCode 2, which excels on this task. We estimate it performs better than 85% of participants on 12 recent Codeforces contests. ↓

GoogleDeepMind's profile picture

Google DeepMind

1,465,405 subscribers

293,039 görüntüleme • 2 yıl önce •via X (Twitter)

Bilim & TeknolojiEğitim
LIVE

Anya Rossi• Live Now

Private livecam show

13 Yorum

Google DeepMind profil fotoğrafı
Google DeepMind2 yıl önce

We’re excited to announce 𝗚𝗲𝗺𝗶𝗻𝗶: @Google’s largest and most capable AI model. Built to be natively multimodal, it can understand and operate across text, code, audio, image and video - and achieves state-of-the-art performance across many tasks. 🧵

Google DeepMind profil fotoğrafı
Google DeepMind2 yıl önce

We’ve optimized Gemini 1.0 for three different sizes, meaning it can run on everything from data centers to mobile phones. 🔨 1️⃣ Ultra: our largest one for highly complex tasks 2️⃣ Pro: our best one for scaling across many tasks 3️⃣ Nano: our most efficient one for devices

Google DeepMind profil fotoğrafı
Google DeepMind2 yıl önce

Gemini Ultra outperforms human experts on MMLU (massive multitask language understanding): one of the most popular methods of benchmarking AI models. It involves a combination of 57 test subjects from math to history to law and more. ↓

Google DeepMind profil fotoğrafı
Google DeepMind2 yıl önce

Gemini is also one of the world’s leading foundation models for coding. It can understand, explain, and generate high-quality code in programming languages like Python, Java, C++, and Go.

Google DeepMind profil fotoğrafı
Google DeepMind2 yıl önce

As Gemini is natively multimodal, it better understands nuanced information and can answer questions relating to complicated topics. This makes it especially good at explaining reasoning in subjects like math and physics - which could be helpful when it comes to homework. ↓

Alexandre Becker profil fotoğrafı
Alexandre Becker2 yıl önce

@RemiLeblond @codeforces Congrats @RemiLeblond 👏

aptr322 profil fotoğrafı
aptr3222 yıl önce

@codeforces are there api or how it's reachable?

Shashwat Srivastava profil fotoğrafı
Shashwat Srivastava2 yıl önce

@codeforces Sound's alarming 😬

Thao Nguyen🇻🇳 profil fotoğrafı
Thao Nguyen🇻🇳2 yıl önce

Using a specialized version of Gemini, THEY created Alpha code 2, which excels on this task

Anonidk2003 profil fotoğrafı
Anonidk20032 yıl önce

@codeforces now thats some hyped up stuff, ask it to participate in a real contest

Vipul Chaudhary profil fotoğrafı
Vipul Chaudhary2 yıl önce

@codeforces These 15% will save earth from these AIs

Azz profil fotoğrafı
Azz2 yıl önce

@codeforces Gimana tanggapannya nih gan @DeeGeeDow

Heaton profil fotoğrafı
Heaton2 yıl önce

@codeforces

Benzer Videolar

The Great Programmer Layoffs Are Coming Google’s new AlphaCode 2 now beats 85% of human programmers. Last year’s AlphaCode was median human level. Yes, in just 1 year, AI went from outperforming 46% to 85% of coders. Sit with that. And it’s actually crazier than that, because for cost reasons they didn’t even use Gemini Ultra, they used the weaker Gemini Pro (~ChatGPT-3.5 ish). How powerful would Gemini Ultra be? Also interesting: Google DeepMind CEO Demis Hassabis said they are also working on research along similar lines to Q*. Sam Altman confirmed Q* was an "unfortunate leak" in a recent interview. Junior devs will likely be the first to go, seniors will have more time.
4:24

The Great Programmer Layoffs Are Coming Google’s new AlphaCode 2 now beats 85% of human programmers. Last year’s AlphaCode was median human level. Yes, in just 1 year, AI went from outperforming 46% to 85% of coders. Sit with that. And it’s actually crazier than that, because for cost reasons they didn’t even use Gemini Ultra, they used the weaker Gemini Pro (~ChatGPT-3.5 ish). How powerful would Gemini Ultra be? Also interesting: Google DeepMind CEO Demis Hassabis said they are also working on research along similar lines to Q*. Sam Altman confirmed Q* was an "unfortunate leak" in a recent interview. Junior devs will likely be the first to go, seniors will have more time.

AI Notkilleveryoneism Memes ⏸️

299,083 görüntüleme • 2 yıl önce

AI that had human-level intelligence would actually be way above human level in capability. Obviously, if we trained a human-level AI, we could just run way more instances of it in parallel. This is a huge advantage. But it's not the only one. Right now, LLMs are much less sample-efficient than humans, but they can process each individual piece of data much more quickly. A model that was a human-level learner would be able to gain the equivalent of thousands of years of education. AIs can also be way more task-oriented than humans. Even a really well-motivated human worker can only operate with ‘human-level’ effort for relatively short spans of time. AGI could do it continuously.
1:28

AI that had human-level intelligence would actually be way above human level in capability. Obviously, if we trained a human-level AI, we could just run way more instances of it in parallel. This is a huge advantage. But it's not the only one. Right now, LLMs are much less sample-efficient than humans, but they can process each individual piece of data much more quickly. A model that was a human-level learner would be able to gain the equivalent of thousands of years of education. AIs can also be way more task-oriented than humans. Even a really well-motivated human worker can only operate with ‘human-level’ effort for relatively short spans of time. AGI could do it continuously.

Dwarkesh Patel

31,861 görüntüleme • 3 ay önce

We also shared evals on Open AI o3-mini — a faster, distilled version of o3 which is optimized for coding, and the first version of o3 we expect to make available for use in early 2025.
4:29

We also shared evals on Open AI o3-mini — a faster, distilled version of o3 which is optimized for coding, and the first version of o3 we expect to make available for use in early 2025.

OpenAI

340,191 görüntüleme • 1 yıl önce

This leaked 2nd Ye verse on Back To Me is wayyyy better than the one we got. Someone has to use AI to make a High Quality version of it ASAP⁉️⁉️
1:09

This leaked 2nd Ye verse on Back To Me is wayyyy better than the one we got. Someone has to use AI to make a High Quality version of it ASAP⁉️⁉️

yzystatus

102,957 görüntüleme • 2 yıl önce

Gemini 3 is truly a great model. It created the Rubik's cube simulation on the first try, in seconds, with ~300 lines of code. Flawless.
0:27

Gemini 3 is truly a great model. It created the Rubik's cube simulation on the first try, in seconds, with ~300 lines of code. Flawless.

Matthew Berman

170,309 görüntüleme • 8 ay önce

Claude Code performs a complete analysis of Gemini-CLI and OpenAI Codex using Deep Graph MCP... one command to implement Claude Code automatically pulls two popular AI assistant repos and performs deep architectural analysis to determine which one is actually better built. Single command to implement: claude mcp add "Deep Graph MCP" npx -- -y mcp-code-graph@latest google-gemini/gemini-cli openai/codex The results might surprise you 👇
1:23

Claude Code performs a complete analysis of Gemini-CLI and OpenAI Codex using Deep Graph MCP... one command to implement Claude Code automatically pulls two popular AI assistant repos and performs deep architectural analysis to determine which one is actually better built. Single command to implement: claude mcp add "Deep Graph MCP" npx -- -y mcp-code-graph@latest google-gemini/gemini-cli openai/codex The results might surprise you 👇

Daniel San

68,001 görüntüleme • 1 yıl önce

Revolutionizing Move Programming with OpenLedger In this demo, we showcase how Move datasets contributed by data providers to OpenLedger’s datanets are used to fine-tune specialized models with LoRA fine-tuning. As seen in the video, we showcase an example on how builders can deploy a Move-specialized model that powers Co-pilot agents using our no-code model fine-tuning platform. This is the future of AI and Web3 innovation. Watch this space to see more specialised models and data feeds being built for next generation agents on top of OpenLedger #Move
3:58

Revolutionizing Move Programming with OpenLedger In this demo, we showcase how Move datasets contributed by data providers to OpenLedger’s datanets are used to fine-tune specialized models with LoRA fine-tuning. As seen in the video, we showcase an example on how builders can deploy a Move-specialized model that powers Co-pilot agents using our no-code model fine-tuning platform. This is the future of AI and Web3 innovation. Watch this space to see more specialised models and data feeds being built for next generation agents on top of OpenLedger #Move

OpenLedger

61,662 görüntüleme • 1 yıl önce

Today's update of Gemini 2.5 Pro Preview is amazing at code web dev, editing, and transformation -- it is now #1 on the WebDev arena, with an ELO score of 1420. That means you can build super cool, visually stunning web apps, with just a single prompt! Check out this example, where we explore the world using this app we created in Gemini's Canvas using the Maps API.
0:30

Today's update of Gemini 2.5 Pro Preview is amazing at code web dev, editing, and transformation -- it is now #1 on the WebDev arena, with an ELO score of 1420. That means you can build super cool, visually stunning web apps, with just a single prompt! Check out this example, where we explore the world using this app we created in Gemini's Canvas using the Maps API.

Tulsee Doshi

115,545 görüntüleme • 1 yıl önce

LEADERSHIP CHANGES EVERYTHING..We will bring Zimbabwe back to better than this level of self sufficiency, a booming economy and reclaiming of a jewel status in the world! We will be first in many things. #NewDirection 
0:59

LEADERSHIP CHANGES EVERYTHING..We will bring Zimbabwe back to better than this level of self sufficiency, a booming economy and reclaiming of a jewel status in the world! We will be first in many things. #NewDirection 

nelson chamisa

74,848 görüntüleme • 2 yıl önce

First Lady Melania Trump delivers remarks at the White House Task Force meeting on AI Education: "We are living in a moment of wonder, and it is our responsibility to prepare children in America."
0:55

First Lady Melania Trump delivers remarks at the White House Task Force meeting on AI Education: "We are living in a moment of wonder, and it is our responsibility to prepare children in America."

Trump War Room

27,593 görüntüleme • 11 ay önce

Jon Cryer: I have Trump supporters in my family—I feel for them on some level, because we in the entertainment industry created this guy. We built up this image of a billionaire, this brilliant businessman, and it was never him. He was always kind of a moron. He still doesn’t understand how tariffs work, even at the most basic level.
0:49

Jon Cryer: I have Trump supporters in my family—I feel for them on some level, because we in the entertainment industry created this guy. We built up this image of a billionaire, this brilliant businessman, and it was never him. He was always kind of a moron. He still doesn’t understand how tariffs work, even at the most basic level.

Acyn

471,876 görüntüleme • 1 yıl önce

President Zelenskyy: The task of Ukrainian units is to inflict a level of losses on the occupier at which Russian casualties exceed the amount of reinforcements they are able to send to their forces each month. This is a realistic task. When we speak of 50,000 Russian losses per month, that is the optimal level. It is a difficult task, without question, but it is precisely the level needed for Russia to begin weighing what it is doing and what it is fighting for. The task of the Ministry of Defense of Ukraine, our army, and all the Defense and Security Forces of Ukraine is to ensure exactly this level of Russian losses.
0:50

President Zelenskyy: The task of Ukrainian units is to inflict a level of losses on the occupier at which Russian casualties exceed the amount of reinforcements they are able to send to their forces each month. This is a realistic task. When we speak of 50,000 Russian losses per month, that is the optimal level. It is a difficult task, without question, but it is precisely the level needed for Russia to begin weighing what it is doing and what it is fighting for. The task of the Ministry of Defense of Ukraine, our army, and all the Defense and Security Forces of Ukraine is to ensure exactly this level of Russian losses.

Anton Gerashchenko

77,058 görüntüleme • 6 ay önce

Quality of this footage is what it is but truly we deserve more concerts in HD from the first leg of the dangerous tour, it was something out of this world, he was simply beaming with energy on stage it was so good, this in my humble opinion is even better than the one we got in Oslo, this was in Monza in 1992, a stunner.
6:02

Quality of this footage is what it is but truly we deserve more concerts in HD from the first leg of the dangerous tour, it was something out of this world, he was simply beaming with energy on stage it was so good, this in my humble opinion is even better than the one we got in Oslo, this was in Monza in 1992, a stunner.

rita

14,468 görüntüleme • 12 gün önce

30 creatives. 34 artisanal AI devices. We called this collection Primavera. Italian for spring - the first green of a more expressive, colorful, human future of AI. At Era, we believe the most meaningful futures are co-created. 🧵🌱
1:16

30 creatives. 34 artisanal AI devices. We called this collection Primavera. Italian for spring - the first green of a more expressive, colorful, human future of AI. At Era, we believe the most meaningful futures are co-created. 🧵🌱

Liz Dorman

334,300 görüntüleme • 3 ay önce

Better/faster/cheaper voice AI turn detection with Gemini 2.0 The code that determines when the agent should respond to the user is some of the most important code in your voice AI agent. The technical terms for this job are "turn detection" or "phrase endpointing." If the voice AI responds before the user has finished their thought, the conversation is choppy and unproductive. If the AI waits too long, the conversation is slow and frustrating. There are a number of ways to approach this. You can: 1. Use a fast "voice activity detection" model to detect pauses in speech. Respond when the user pauses. 2. Use a specialized phrase endpointing model that operates on transcribed text, pattern-matching on text semantics. 3. Train a specialized phrase endpointing model that operates directly on audio. 4. Leverage the native audio capabilities of a SOTA LLM like Gemini 2.0. We've benchmarked all four of these, and Gemini 2.0 currently beats other approaches. Using Gemini is also cheaper than transcribing the audio separately using a transcription service or model. Here's a short video showing Gemini phrase endpointing in two scenarios. First, correctly handling pauses in natural conversation. Second, requesting a phone number (which is a common activity in a use case like customer support). You can see the Completeness check lines in the terminal output, printed each time Gemini processes a chunk of audio.
1:23

Better/faster/cheaper voice AI turn detection with Gemini 2.0 The code that determines when the agent should respond to the user is some of the most important code in your voice AI agent. The technical terms for this job are "turn detection" or "phrase endpointing." If the voice AI responds before the user has finished their thought, the conversation is choppy and unproductive. If the AI waits too long, the conversation is slow and frustrating. There are a number of ways to approach this. You can: 1. Use a fast "voice activity detection" model to detect pauses in speech. Respond when the user pauses. 2. Use a specialized phrase endpointing model that operates on transcribed text, pattern-matching on text semantics. 3. Train a specialized phrase endpointing model that operates directly on audio. 4. Leverage the native audio capabilities of a SOTA LLM like Gemini 2.0. We've benchmarked all four of these, and Gemini 2.0 currently beats other approaches. Using Gemini is also cheaper than transcribing the audio separately using a transcription service or model. Here's a short video showing Gemini phrase endpointing in two scenarios. First, correctly handling pauses in natural conversation. Second, requesting a phone number (which is a common activity in a use case like customer support). You can see the Completeness check lines in the terminal output, printed each time Gemini processes a chunk of audio.

kwindla

64,188 görüntüleme • 1 yıl önce

Google vibe coded so hard we can now code frontend by drawing on it like a whiteboard 😂 While you're in Google AI Studio, just grab the drawing tools, mark up any UI element, and watch Gemini turn your scribbles into actual code. oh yea now we don't write code anymore, we draw it
0:45

Google vibe coded so hard we can now code frontend by drawing on it like a whiteboard 😂 While you're in Google AI Studio, just grab the drawing tools, mark up any UI element, and watch Gemini turn your scribbles into actual code. oh yea now we don't write code anymore, we draw it

corbin

187,730 görüntüleme • 9 ay önce

To my knowledge, this is the first publicly available precinct-level estimate of Presidential support by demographic on a national scale. Here’s a quick demo of how it works:
1:24

To my knowledge, this is the first publicly available precinct-level estimate of Presidential support by demographic on a national scale. Here’s a quick demo of how it works:

Zachary Donnini

1,729,670 görüntüleme • 10 ay önce

We gotta respect 🐐s of this shit man this version of lil Wayne prob the best rapper ever .. only ppl on this level or better was possibly Pac maybe a 2015-2016 Future
4:25

We gotta respect 🐐s of this shit man this version of lil Wayne prob the best rapper ever .. only ppl on this level or better was possibly Pac maybe a 2015-2016 Future

Jaqen H🐗'ghar

13,171 görüntüleme • 5 ay önce

When we were first putting together this level, a bug caused one of the mushroom NPCs to massively size up and it was hilarious. So much so, that we decided to write it into the game. Happy game dev accidents.
0:29

When we were first putting together this level, a bug caused one of the mushroom NPCs to massively size up and it was hilarious. So much so, that we decided to write it into the game. Happy game dev accidents.

EPG ➡️ Teeto out July 15th!

44,869 görüntüleme • 3 ay önce

Kira (Short Film on Human Cloning) My new AI-assisted short film is here. Kira explores human cloning and the search for identity in today’s world. It took nearly 600 prompts, 12 days (during my free time), and a $500 budget to bring this project to life. The entire film was created by one person using a range of AI tools, all listed at the end. Enjoy. ~ Hashem
15:57

Kira (Short Film on Human Cloning) My new AI-assisted short film is here. Kira explores human cloning and the search for identity in today’s world. It took nearly 600 prompts, 12 days (during my free time), and a $500 budget to bring this project to life. The entire film was created by one person using a range of AI tools, all listed at the end. Enjoy. ~ Hashem

Hashem Al-Ghaili

101,015 görüntüleme • 1 yıl önce