Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Amazon’s machine learning model collects 300 million data points per season and can now predict which players are likely to blitz before the snap. This is a look into the future of broadcasting.

JoePompliano's profile picture

Joe Pompliano

605,426 subscribers

688,472 görüntüleme • 2 yıl önce •via X (Twitter)

Bilim & TeknolojiHaberler & PolitikaSpor
LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

Amazon’s machine learning model is so cool. It highlights both potential blitzes before the play, Alvin Kamara doesn’t pick it up, and Derek Carr gets hit as he throws. It's only a matter of time before teams want access to this data.
0:17

Amazon’s machine learning model is so cool. It highlights both potential blitzes before the play, Alvin Kamara doesn’t pick it up, and Derek Carr gets hit as he throws. It's only a matter of time before teams want access to this data.

Joe Pompliano

12,159,083 görüntüleme • 2 yıl önce

All of Us has significantly expanded our data to now include nearly a quarter million whole genome sequences! Researchers can use this data to make scientific discoveries that can lead to a healthier future for everyone. Read the announcement #JoinAllofUs
0:15

All of Us has significantly expanded our data to now include nearly a quarter million whole genome sequences! Researchers can use this data to make scientific discoveries that can lead to a healthier future for everyone. Read the announcement #JoinAllofUs

AllofUsResearch

36,535 görüntüleme • 3 yıl önce

The future is now: new paper where we developed & validated the first commercially available wearable to predict tissue loading at three common running injury locations using machine learning Full-text Free formatted full-text
0:10

The future is now: new paper where we developed & validated the first commercially available wearable to predict tissue loading at three common running injury locations using machine learning Full-text Free formatted full-text

Bas Van Hooren

32,580 görüntüleme • 2 yıl önce

This is what the future of football will look like, at least in the teams I coach. A mix of positional players- who give the team depth & support under the ball and functional players who are free to migrate to the ball zone. Moving actions aren’t exclusive to functional players
0:13

This is what the future of football will look like, at least in the teams I coach. A mix of positional players- who give the team depth & support under the ball and functional players who are free to migrate to the ball zone. Moving actions aren’t exclusive to functional players

JOGO FUNCIONAL

11,297 görüntüleme • 1 yıl önce

£1 million per week in wages and this is what he has delivered in 2026. He is one of the reasons we are having such a poor season.
0:13

£1 million per week in wages and this is what he has delivered in 2026. He is one of the reasons we are having such a poor season.

City Chief

609,202 görüntüleme • 5 ay önce

In flow matching, a coupling determines how noise and data samples are paired during training. The choice of coupling is important because it influences the geometry of trajectories at inference time. The simplest choice is the independent coupling, where noise and data points are paired arbitrarily. This can lead to curved trajectories as the model averages over many conflicting pairings. However, if we use optimal transport on batches of pairs, this leads to fewer ambiguous intersections that the model must resolve, leading to straighter trajectories at inference time.
0:18

In flow matching, a coupling determines how noise and data samples are paired during training. The choice of coupling is important because it influences the geometry of trajectories at inference time. The simplest choice is the independent coupling, where noise and data points are paired arbitrarily. This can lead to curved trajectories as the model averages over many conflicting pairings. However, if we use optimal transport on batches of pairs, this leads to fewer ambiguous intersections that the model must resolve, leading to straighter trajectories at inference time.

Alec Helbling

65,484 görüntüleme • 3 ay önce

Predicting how the environment changes over time is a key task in robotics, enabling all kinds of downstream tasks. Trace Anything is a very interesting look at how you can train models which predict the trajectory each point will take in the future -- you can use this for robot manipulation, video prediction, and more. Learn more ->
0:16

Predicting how the environment changes over time is a key task in robotics, enabling all kinds of downstream tasks. Trace Anything is a very interesting look at how you can train models which predict the trajectory each point will take in the future -- you can use this for robot manipulation, video prediction, and more. Learn more ->

Chris Paxton

56,845 görüntüleme • 8 ay önce

Added context to my tiny diffusion model to enable sequential generation of longer outputs! Currently the context is a quarter of the sequence length (seq_len=256, context_len=64). I have a theory that the less semantic-value-per-token, the worse the “curse of parallel decoding” is. With parallel decoding, we independently predict multiple tokens in one step. With the sentence “My poker hand was a ___ ___”, two valid predictions are “two pair” and “straight flush”. Because each token prediction is independent though, we can end up with a nonsensical output like “two flush”. This seems to be exacerbated with low semantic-value-per-token, as now you need more tokens to express the same concept. Instead of needing to independently predict two tokens, we might need to predict 10 instead (which is of course much harder). The model currently has noticeably worse output compared to nanogpt (similar size) and I believe this is a main reason. I’ll try adding confidence-aware parallel decoding (from NVIDIA’s Fast-dLLM paper) and other tricks and see how much they improve generation quality.
0:12

Added context to my tiny diffusion model to enable sequential generation of longer outputs! Currently the context is a quarter of the sequence length (seq_len=256, context_len=64). I have a theory that the less semantic-value-per-token, the worse the “curse of parallel decoding” is. With parallel decoding, we independently predict multiple tokens in one step. With the sentence “My poker hand was a ___ ___”, two valid predictions are “two pair” and “straight flush”. Because each token prediction is independent though, we can end up with a nonsensical output like “two flush”. This seems to be exacerbated with low semantic-value-per-token, as now you need more tokens to express the same concept. Instead of needing to independently predict two tokens, we might need to predict 10 instead (which is of course much harder). The model currently has noticeably worse output compared to nanogpt (similar size) and I believe this is a main reason. I’ll try adding confidence-aware parallel decoding (from NVIDIA’s Fast-dLLM paper) and other tricks and see how much they improve generation quality.

Nathan Barry

89,040 görüntüleme • 10 ay önce

🚨DATA: KASPA L1 APPROACHES 600 MILLION TRANSACTIONS The Kaspa L1 network is now just inches from reaching the 600 million cumulative transaction milestone, according to Kaspa Explorer. $KAS' mainnet was launches back in 2021 and, per conversations we have had with key stakeholders, 2026 is set to be a year of intense building. If you're looking to get involved in Kaspa this year, take a look at our comprehensive look into the best wallet platforms that support $KAS ⬇️
0:18

🚨DATA: KASPA L1 APPROACHES 600 MILLION TRANSACTIONS The Kaspa L1 network is now just inches from reaching the 600 million cumulative transaction milestone, according to Kaspa Explorer. $KAS' mainnet was launches back in 2021 and, per conversations we have had with key stakeholders, 2026 is set to be a year of intense building. If you're looking to get involved in Kaspa this year, take a look at our comprehensive look into the best wallet platforms that support $KAS ⬇️

BSCN

24,237 görüntüleme • 6 ay önce

Dear friends, we are excited to introduce Data Points, a rewards system built on the legacy of the RAS Factor System. Data Points reward engagement within the Dataland ecosystem—bridging past contributions with future benefits in both digital and physical realms.
0:18

Dear friends, we are excited to introduce Data Points, a rewards system built on the legacy of the RAS Factor System. Data Points reward engagement within the Dataland ecosystem—bridging past contributions with future benefits in both digital and physical realms.

Refik Anadol

18,749 görüntüleme • 1 yıl önce

TON-NATIVE MEMECOINS REACH $156 MILLION MARKET CAP According to CoinGecko data, the combined market cap of memecoins hosted on the TON 💎 has surged to a value of nearly $157 million. The $TON token itself has rallied alongside and the recent listings of both $NOT and $DOGS by leading neobank Revolut has likely played a role in reviving the ecosystem. Which TON memecoin is set to explode this year...?
0:16

TON-NATIVE MEMECOINS REACH $156 MILLION MARKET CAP According to CoinGecko data, the combined market cap of memecoins hosted on the TON 💎 has surged to a value of nearly $157 million. The $TON token itself has rallied alongside and the recent listings of both $NOT and $DOGS by leading neobank Revolut has likely played a role in reviving the ecosystem. Which TON memecoin is set to explode this year...?

BSCN

10,388 görüntüleme • 3 ay önce

Justin Fields torched the Cowboys for 25.9 fantasy points in Week 5. Most of which was in garbage time. Week 6? He faces the Broncos...third-fewest fantasy points and rushing yards per game allowed to QBs this season. Consider pivoting to a streamer this week.
0:15

Justin Fields torched the Cowboys for 25.9 fantasy points in Week 5. Most of which was in garbage time. Week 6? He faces the Broncos...third-fewest fantasy points and rushing yards per game allowed to QBs this season. Consider pivoting to a streamer this week.

Moody

27,685 görüntüleme • 10 ay önce

Some players are out of the starting blocks early this season. Players like Scrzbbles have been on the grind in a major way heading into the RLCS season. Also, Chausette?!!
0:14

Some players are out of the starting blocks early this season. Players like Scrzbbles have been on the grind in a major way heading into the RLCS season. Also, Chausette?!!

belleair

13,509 görüntüleme • 1 yıl önce

🚨 ICYMI: Nansen Points Season 02 went live yesterday with the launch of Nansen AI mobile. This season: New perks and new ways to turn your Points into real rewards. Here’s a look at what’s waiting for you 👇
0:16

🚨 ICYMI: Nansen Points Season 02 went live yesterday with the launch of Nansen AI mobile. This season: New perks and new ways to turn your Points into real rewards. Here’s a look at what’s waiting for you 👇

Nansen 🧭

85,398 görüntüleme • 11 ay önce

The future of gaming is here. 🚀 BreederDAO is evolving into Sovrun, and we’re stepping into the era of Autonomous Worlds—where players, creators, and communities govern and build in ways never seen before.
0:17

The future of gaming is here. 🚀 BreederDAO is evolving into Sovrun, and we’re stepping into the era of Autonomous Worlds—where players, creators, and communities govern and build in ways never seen before.

Sovrun

59,701 görüntüleme • 1 yıl önce

Coming soon: @Tan_Stack Router's `intent` preloading can now predict where your users cursor will go (you can even customize proximity) to preload the routes that users are most likely to visit *before* they even hover the link. Prefetch smarter, not harder.
0:11

Coming soon: @Tan_Stack Router's `intent` preloading can now predict where your users cursor will go (you can even customize proximity) to preload the routes that users are most likely to visit *before* they even hover the link. Prefetch smarter, not harder.

Tanner Linsley

140,486 görüntüleme • 1 yıl önce

Flare is an EVM smart contract platform specially designed to support data-intensive use cases, including machine learning and real-world assets. Explore what Flare ☀️, the Blockchain for Data, can offer. 🔗
0:15

Flare is an EVM smart contract platform specially designed to support data-intensive use cases, including machine learning and real-world assets. Explore what Flare ☀️, the Blockchain for Data, can offer. 🔗

Flare ☀️

34,331 görüntüleme • 2 yıl önce