https://t.co/jq2TzJMrHT
We’re here to help people/orgs choose the best model and provider for real AI use cases.
Should you be using GPT-4 on Azure or Mixtral on Perplexity? We’ll help you decide.
A few highlights: 🧵
There’s a trade-off between model quality and throughput (speed), with higher quality models typically having lower throughput (slower output).
If you need maximum quality, GPT-4 is still the only game in town (watch this space though!).
Comparing between models: Quality vs. Throughput
There’s an order of magnitude difference in pricing between models.
If you don’t need the quality of GPT-4, there are several competitive options including Gemini Pro, Mixtral 8x7B and GPT-3.5 Turbo.
We’re here to help people/orgs choose the best model and provider for real AI use cases.
Should you be using GPT-4 on Azure or Mixtral on Perplexity? We’ll help you decide.
A few highlights: 🧵
There’s a trade-off between model quality and throughput (speed), with higher quality models typically having lower throughput (slower output).
If you need maximum quality, GPT-4 is still the only game in town (watch this space though!).
Comparing between models: Quality vs. Throughput
There’s an order of magnitude difference in pricing between models.
If you don’t need the quality of GPT-4, there are several competitive options including Gemini Pro, Mixtral 8x7B and GPT-3.5 Turbo.
Forwarded from 전종현의 인사이트
그리고 한가지 감사한 일이 있어서 공유하려고 합니다. 얼마전에 제 AI 리서치 클럽 2기 오프라인 모임이 있었는데, 감사하게도 크라이치즈버거에서 햄버거를 무상으로 지원해주셨습니다. 편지까지 직접 작성해서 보내주셨는데 내용에 진심이 담겨있어서 다시 한번 감동이었네요.
모두들 크라이치즈버거 많이 사랑해주세요 👍
”사업이 잘 되었으면 좋겠습니다. 돈을 벌고 싶어서요? 그것도 맞습니다. 돈을 못 벌면 안 되지요. 그런데 더 중요한 이유가 있습니다. 따뜻한 햄버거와, 따뜻한 음식점을 만들고 싶었습니다. 동양인 혼자 영어를 못하면서 쩔쩔매며 버거를 먹어도, 따뜻한 마음으로 인간적으로 대해주는 그런 음식점, 공간을 만들고 싶었습니다.”
”거대형 프랜차이즈가 아니지만.. 성실함과, 진정성으로 '천천히, 오래' 꾸준히 사업을 해서 성공할 수 있다는 것을 살면서 증명해 보고 싶고요. 그리고 무엇보다.. 저와 함께 해주시는 직원분들이 저의 꿈을 믿고 함께 해주신 것에 대해서 절대 후회 없는 시간이라고 생각하게 해주고 싶어서요. 진심입니다.”
https://www.daangn.com/kr/business-posts/65535b79abc33439f564225d
모두들 크라이치즈버거 많이 사랑해주세요 👍
”사업이 잘 되었으면 좋겠습니다. 돈을 벌고 싶어서요? 그것도 맞습니다. 돈을 못 벌면 안 되지요. 그런데 더 중요한 이유가 있습니다. 따뜻한 햄버거와, 따뜻한 음식점을 만들고 싶었습니다. 동양인 혼자 영어를 못하면서 쩔쩔매며 버거를 먹어도, 따뜻한 마음으로 인간적으로 대해주는 그런 음식점, 공간을 만들고 싶었습니다.”
”거대형 프랜차이즈가 아니지만.. 성실함과, 진정성으로 '천천히, 오래' 꾸준히 사업을 해서 성공할 수 있다는 것을 살면서 증명해 보고 싶고요. 그리고 무엇보다.. 저와 함께 해주시는 직원분들이 저의 꿈을 믿고 함께 해주신 것에 대해서 절대 후회 없는 시간이라고 생각하게 해주고 싶어서요. 진심입니다.”
https://www.daangn.com/kr/business-posts/65535b79abc33439f564225d
Nvidia CEO says countries need own AI infrastructure
ARTIFICIAL INTELLIGENCE: Jensen Huang said fears about the dangers of AI are overblown, adding every country should build its own infrastructure as fast as it can
Reuters and Bloomberg, AMSTERDAM and SYDNEY
Nvidia Corp chief executive officer Jensen Huang (黃仁勳) said yesterday that every country needs to have its own artificial intelligence (AI) infrastructure to take advantage of the economic potential while protecting its own culture.
“You cannot allow that to be done by other people,” Huang said at the World Government Summit in Dubai.
Huang, whose firm has catapulted to a US$1.73 trillion stock market value due to its dominance of the market for high-end AI chips, said his company is “democratizing” access to AI due to swift efficiency gains in AI computing.
Nvidia CEO Jensen Huang attends a session of the World Governments Summit in Dubai, United Arab Emirates, yesterday.
Photo: Reuters
“The rest of it is really up to you to take initiative, activate your industry, build the infrastructure, as fast as you can,” he said.
He said that fears about the dangers of AI are overblown, adding that other new technologies and industries such as cars and aviation have been successfully regulated.
“There are some interests to scare people about this new technology, to mystify this technology, to encourage other people to not do anything about that technology and rely on them to do it. And I think that’s a mistake,” Huang said.
Huang anticipates advances in computing over the next few years will keep the cost of developing AI well below the US$7 trillion that Sam Altman is said to be fundraising.
“You can’t assume just that you will buy more computers. You have to also assume that the computers are going to become faster and therefore the total amount that you need is not as much,” Huang told the summit.
In his remarks, Huang estimated that the global cost of data centers powering AI will double in the next five years.
“We’re at the beginning of this new era. There’s about a trillion dollars’ worth of installed base of data centers. Over the course of the next four or five years, we’ll have US$2 trillion worth of data centers that will be powering software around the world,” he said.
ARTIFICIAL INTELLIGENCE: Jensen Huang said fears about the dangers of AI are overblown, adding every country should build its own infrastructure as fast as it can
Reuters and Bloomberg, AMSTERDAM and SYDNEY
Nvidia Corp chief executive officer Jensen Huang (黃仁勳) said yesterday that every country needs to have its own artificial intelligence (AI) infrastructure to take advantage of the economic potential while protecting its own culture.
“You cannot allow that to be done by other people,” Huang said at the World Government Summit in Dubai.
Huang, whose firm has catapulted to a US$1.73 trillion stock market value due to its dominance of the market for high-end AI chips, said his company is “democratizing” access to AI due to swift efficiency gains in AI computing.
Nvidia CEO Jensen Huang attends a session of the World Governments Summit in Dubai, United Arab Emirates, yesterday.
Photo: Reuters
“The rest of it is really up to you to take initiative, activate your industry, build the infrastructure, as fast as you can,” he said.
He said that fears about the dangers of AI are overblown, adding that other new technologies and industries such as cars and aviation have been successfully regulated.
“There are some interests to scare people about this new technology, to mystify this technology, to encourage other people to not do anything about that technology and rely on them to do it. And I think that’s a mistake,” Huang said.
Huang anticipates advances in computing over the next few years will keep the cost of developing AI well below the US$7 trillion that Sam Altman is said to be fundraising.
“You can’t assume just that you will buy more computers. You have to also assume that the computers are going to become faster and therefore the total amount that you need is not as much,” Huang told the summit.
In his remarks, Huang estimated that the global cost of data centers powering AI will double in the next five years.
“We’re at the beginning of this new era. There’s about a trillion dollars’ worth of installed base of data centers. Over the course of the next four or five years, we’ll have US$2 trillion worth of data centers that will be powering software around the world,” he said.
Forwarded from Nikkei Asia
India's Modi woos UAE and Qatar to counter China in Middle East
Indian Prime Minister Narendra Modi has embarked on a trip to the United Arab Emirates and Qatar, seeking to increase economic cooperation and bolster ties in the Middle East as China expands its influence there.
Read more here
Indian Prime Minister Narendra Modi has embarked on a trip to the United Arab Emirates and Qatar, seeking to increase economic cooperation and bolster ties in the Middle East as China expands its influence there.
Read more here
Forwarded from 요즘AI
This media is not supported in your browser
VIEW IN TELEGRAM
오픈AI에서 Text-to-Video 모델인 Sora를 공개했습니다
기존의 DALL.E모델과 GPT 연구를 기반으로 만들어졌다고 합니다.
아직 모델이 완벽하지는 않다고 하지만 공개된 영상들의 퀄리티는 미친 수준이네요..
https://openai.com/sora#research
기존의 DALL.E모델과 GPT 연구를 기반으로 만들어졌다고 합니다.
아직 모델이 완벽하지는 않다고 하지만 공개된 영상들의 퀄리티는 미친 수준이네요..
https://openai.com/sora#research
Forwarded from 요즘AI
https://blog.google/intl/ko-kr/products/explore-get-answers/google-gemini-next-generation-model-february-2024-kr/
오픈AI의 Sora가 오늘의 이슈였는데, 구글의 Gemini도 1.5버전을 공개했습니다!
- 무려 100만 토큰을 지원(현재 GPT의 토큰 한도: 32k)
- 트랜스포머 + MoE 구조
오픈AI의 Sora가 오늘의 이슈였는데, 구글의 Gemini도 1.5버전을 공개했습니다!
- 무려 100만 토큰을 지원(현재 GPT의 토큰 한도: 32k)
- 트랜스포머 + MoE 구조
If you think OpenAI Sora is a creative toy like DALLE, ... think again. Sora is a data-driven physics engine. It is a simulation of many worlds, real or fantastical. The simulator learns intricate rendering, "intuitive" physics, and long-horizon consistency, all by some denoising and gradient maths.
I won't be surprised if Sora is trained on lots of synthetic data using Unreal Engine 5. It has to be!
Let's breakdown the following video. Prompt: "Photorealistic closeup video of two pirate ships battling each other as they sail inside a cup of coffee."
- The simulator instantiates two exquisite 3D assets: pirate ships with different decorations. Sora has to solve text-to-3D implicitly in its latent space.
- The 3D objects are consistently animated as they sail and avoid each other's paths.
- Fluid dynamics of the coffee, even the foams that form around the ships. Fluid simulation is an entire sub-field of computer graphics, which traditionally requires very complex algorithms and equations.
- Photorealism, almost like rendering with raytracing.
- The simulator takes into account the small size of the cup compared to oceans, and applies tilt-shift photography to give a "minuscule" vibe.
- The semantics of the scene does not exist in the real world, but the engine still implements the correct physical rules that we expect.
Next up: add more modalities and conditioning, then we have a full data-driven UE that will replace all the hand-engineered graphics pipelines.
Apparently some folks don't get "data-driven physics engine", so let me clarify. Sora is an end-to-end, diffusion transformer model. It inputs text/image and outputs video pixels directly. Sora learns a physics engine implicitly in the neural parameters by gradient descent through massive amounts of videos.
Sora is a learnable simulator, or "world model". Of course it does not call UE5 explicitly in the loop, but it's possible that UE5-generated (text, video) pairs are added as synthetic data to the training set.
I won't be surprised if Sora is trained on lots of synthetic data using Unreal Engine 5. It has to be!
Let's breakdown the following video. Prompt: "Photorealistic closeup video of two pirate ships battling each other as they sail inside a cup of coffee."
- The simulator instantiates two exquisite 3D assets: pirate ships with different decorations. Sora has to solve text-to-3D implicitly in its latent space.
- The 3D objects are consistently animated as they sail and avoid each other's paths.
- Fluid dynamics of the coffee, even the foams that form around the ships. Fluid simulation is an entire sub-field of computer graphics, which traditionally requires very complex algorithms and equations.
- Photorealism, almost like rendering with raytracing.
- The simulator takes into account the small size of the cup compared to oceans, and applies tilt-shift photography to give a "minuscule" vibe.
- The semantics of the scene does not exist in the real world, but the engine still implements the correct physical rules that we expect.
Next up: add more modalities and conditioning, then we have a full data-driven UE that will replace all the hand-engineered graphics pipelines.
Apparently some folks don't get "data-driven physics engine", so let me clarify. Sora is an end-to-end, diffusion transformer model. It inputs text/image and outputs video pixels directly. Sora learns a physics engine implicitly in the neural parameters by gradient descent through massive amounts of videos.
Sora is a learnable simulator, or "world model". Of course it does not call UE5 explicitly in the loop, but it's possible that UE5-generated (text, video) pairs are added as synthetic data to the training set.
❤1