Tree-mounted machines connect to web app to let users knock on wood remotely
71 by surprisetalk | 32 comments on Hacker News.
Jumat, 30 Oktober 2026
Selasa, 27 Oktober 2026
Jalen Hurts' mom's students send well-wishes in QB's road to recovery
Pamela Hurts, a schoolteacher, was clearly touched by the cards from her students on "Jalen Hurts Day."
from www.espn.com - TOP https://ift.tt/auCVjQD
from www.espn.com - TOP https://ift.tt/auCVjQD
Senin, 26 Oktober 2026
New top story on Hacker News: Stalwart v1.0: The Fastest Mail Server on the Planet
Stalwart v1.0: The Fastest Mail Server on the Planet
32 by StalwartLabs | 11 comments on Hacker News.
32 by StalwartLabs | 11 comments on Hacker News.
Time to trade red-hot Ty Jerome, drop struggling Draymond Green?
Ty Jerome's fantasy stock is seemingly on a rocket ship, but should fantasy basketball managers keep the faith in Myles Turner and Onyeka Okongwu?
from www.espn.com - TOP https://ift.tt/nb0zvlB
from www.espn.com - TOP https://ift.tt/nb0zvlB
Minggu, 25 Oktober 2026
Suited and booted: Aaron Rodgers, Steelers QB room lead NFL Week 7 arrivals
NFL stars continue to turn game-day arrivals into their own weekly fashion showcase.
from www.espn.com - TOP https://ift.tt/S1GqDK8
from www.espn.com - TOP https://ift.tt/S1GqDK8
Sabtu, 24 Oktober 2026
UFC 333 live results, analysis for Volkanovski vs. Evloev, Yan vs. Dvalishvili
ESPN MMA contributors break down all the action from UFC 333 in Abu Dhabi.
from www.espn.com - TOP https://ift.tt/7eAJfRD
from www.espn.com - TOP https://ift.tt/7eAJfRD
Jumat, 23 Oktober 2026
Chase Elliott opts out of popular driver award to steer vote to Kyle Busch
Perennial winner Chase Elliott has opted to remove his name from consideration for NASCAR's Most Popular Driver Award out of respect for the late Kyle Busch.
from www.espn.com - TOP https://ift.tt/1tZyaEW
from www.espn.com - TOP https://ift.tt/1tZyaEW
Kamis, 22 Oktober 2026
Locksley 'at peace' with Maryland firing, open to NFL, college OC jobs
Coach Mike Locksley says he's "at peace" with his firing by Maryland, telling ESPN that he's open to jobs in the NFL or offensive coordinator duties at a "championship" school.
from www.espn.com - TOP https://ift.tt/GgDnWMm
from www.espn.com - TOP https://ift.tt/GgDnWMm
UConn's Strong, USC's Watkins headline preseason AP All-American women's team
UConn's Sarah Strong and USC's JuJu Watkins, who have won The Associated Press Player of the Year award the past two years, headline the preseason AP All-America women's college basketball team released Thursday.
from www.espn.com - TOP https://ift.tt/wWe7KZX
from www.espn.com - TOP https://ift.tt/wWe7KZX
Rabu, 21 Oktober 2026
Maryland fires Mike Locksley: Top candidates and roster priorities
After seven full seasons in College Park, Locksley is out, but the Terps could have some notable pieces to build around.
from www.espn.com - TOP https://ift.tt/vimr52e
from www.espn.com - TOP https://ift.tt/vimr52e
Selasa, 20 Oktober 2026
AP preseason Top 25 poll reaction: What to know about every team
The first poll of the 2026-27 season is here. We look at the projected starters and biggest game for each team.
from www.espn.com - TOP https://ift.tt/6cQu13n
from www.espn.com - TOP https://ift.tt/6cQu13n
Senin, 19 Oktober 2026
NWSL: San Diego Wave stays No. 1, Angel City keeps rising
Can anyone stop the San Diego Wave? Their fifth straight win keeps them top of ESPN's NWSL Power Rankings after Matchday 25.
from www.espn.com - TOP https://ift.tt/tpdanSF
from www.espn.com - TOP https://ift.tt/tpdanSF
Minggu, 18 Oktober 2026
Seven games, seven points: Man United are sleepwalking to another lost season
Manchester United were already suffering from their worst start to a Premier League season since 1986. A 1-1 draw with Leeds won't help.
from www.espn.com - TOP https://ift.tt/bomFNZc
from www.espn.com - TOP https://ift.tt/bomFNZc
Sabtu, 17 Oktober 2026
U.S. loses gold medalist Rivera (ankle) for gymnastics worlds
The U.S. women's gymnastics team will have to make a go of it at the world championships without Hezly Rivera.
from www.espn.com - TOP https://ift.tt/vtMbz76
from www.espn.com - TOP https://ift.tt/vtMbz76
Jumat, 16 Oktober 2026
Spanish team challenging Thunder rookie Mara's NBA eligibility
A Spanish basketball team is challenging Thunder rookie Aday Mara's eligibility to play in the NBA this season, saying he's still under contract with the team.
from www.espn.com - TOP https://ift.tt/6U8gwuz
from www.espn.com - TOP https://ift.tt/6U8gwuz
The expansion no one asked for: CFP's 24-team move has problems
The CFP is nearing a deal to make the tournament bigger. Bigger isn't better, and this one could really miss.
from www.espn.com - TOP https://ift.tt/6QTguiI
from www.espn.com - TOP https://ift.tt/6QTguiI
Kamis, 15 Oktober 2026
Amar'e Stoudemire graduates from law enforcement academy, will train as an officer
Stoudemire will now train with the Miami-Dade Schools Police Department.
from www.espn.com - TOP https://ift.tt/SBGIEkN
from www.espn.com - TOP https://ift.tt/SBGIEkN
Rabu, 14 Oktober 2026
🔴 Champions League live updates: Liverpool in action as early matches kick off
Follow live updates, scores and highlights from Wednesday's UEFA Champions League action, featuring Manchester City vs. Paris Saint-Germain, AS Roma vs. Real Madrid and more.
from www.espn.com - TOP https://ift.tt/xFdyIV4
from www.espn.com - TOP https://ift.tt/xFdyIV4
Selasa, 13 Oktober 2026
Why did the Cardinals trade Harrison? And what does he bring to the Texans?
After a 1-4 start, Houston is bringing in Harrison in hopes of giving Stroud another playmaker.
from www.espn.com - TOP https://ift.tt/qhfWuvE
from www.espn.com - TOP https://ift.tt/qhfWuvE
Senin, 12 Oktober 2026
Alabama WR Coleman-Williams unlikely to play vs. Vols, says DeBoer
Alabama star receiver Ryan Coleman-Williams is unlikely to play at Tennessee on Saturday due to multiple injuries, according to coach Kalen DeBoer.
from www.espn.com - TOP https://ift.tt/zyeFToQ
from www.espn.com - TOP https://ift.tt/zyeFToQ
Minggu, 11 Oktober 2026
Boston sports nostalgia headlines NFL Week 5's best arrivals
NFL stars continue to turn game-day arrivals into their own weekly fashion showcase.
from www.espn.com - TOP https://ift.tt/Ag8IRPO
from www.espn.com - TOP https://ift.tt/Ag8IRPO
Sabtu, 10 Oktober 2026
Rangers G Shesterkin out for season with right knee injury
Rangers goaltender Igor Shesterkin will miss the rest of the season after suffering a right knee injury in Friday night's loss to the Capitals.
from www.espn.com - TOP https://ift.tt/zt1n74y
from www.espn.com - TOP https://ift.tt/zt1n74y
Jumat, 09 Oktober 2026
Panthers expecting Marchand (leg) to miss six weeks
The Panthers are expecting forward Brad Marchand to miss about six weeks with his left leg injury, a timeline that coincides with Aleksander Barkov being out six to eight weeks.
from www.espn.com - TOP https://ift.tt/4L1azsr
from www.espn.com - TOP https://ift.tt/4L1azsr
Kamis, 08 Oktober 2026
Commanders OC: Daniels will be 'full go,' have no limitations vs. Giants
Commanders offensive coordinator David Blough said Jayden Daniels will have no limitations when the star QB plays Sunday against the Giants.
from www.espn.com - TOP https://ift.tt/bU1M247
from www.espn.com - TOP https://ift.tt/bU1M247
Rabu, 07 Oktober 2026
New top story on Hacker News: Show HN: A modern low-poly SimCity 2000 clone, built by Opus 5.5
Show HN: A modern low-poly SimCity 2000 clone, built by Opus 5.5
14 by drdator | 3 comments on Hacker News.
This is a clone of simcity 2000 I (using Opus 5.5) built over a few days. It started with a single prompt "create a typescript webgl low-poly version (full replica of game logic) of simcity 2000. the original is available here for inspection [...].zip". It then decompiled it using Ghidra and started translating the old native PowerPC code function-by-function. After about 2 hours, using 27 subagents, i had a running version in my browser. After a follow up prompt (make sure all aspects of the game is implemented correctly and accurately), Opus decided to use a PowerPC emulator (Unicorn) and write a harness with faked Mac system calls in Python for verification, running 10,000 game days generating an identical game state. After that it has basically just been polishing, and adding a few subtle modernizations like, smog, depth-of-field and environmental sounds to make it a bit more immersive without killing the original nostalgic vibe. All 3D and vector assets are generated with code by Opus, and the bitmap image assets are generated by GPT Image 2. The music and sounds are not genai because all AI audio models i've tried sound terrible, so the music and sounds are CC-licensed. The music is from Kevin MacLeod, https://ift.tt/fKMetXs I created this because I wanted it and to see how capable Opus 5.5 was. It turned out much better than expected, so I figured I'd share it.
14 by drdator | 3 comments on Hacker News.
This is a clone of simcity 2000 I (using Opus 5.5) built over a few days. It started with a single prompt "create a typescript webgl low-poly version (full replica of game logic) of simcity 2000. the original is available here for inspection [...].zip". It then decompiled it using Ghidra and started translating the old native PowerPC code function-by-function. After about 2 hours, using 27 subagents, i had a running version in my browser. After a follow up prompt (make sure all aspects of the game is implemented correctly and accurately), Opus decided to use a PowerPC emulator (Unicorn) and write a harness with faked Mac system calls in Python for verification, running 10,000 game days generating an identical game state. After that it has basically just been polishing, and adding a few subtle modernizations like, smog, depth-of-field and environmental sounds to make it a bit more immersive without killing the original nostalgic vibe. All 3D and vector assets are generated with code by Opus, and the bitmap image assets are generated by GPT Image 2. The music and sounds are not genai because all AI audio models i've tried sound terrible, so the music and sounds are CC-licensed. The music is from Kevin MacLeod, https://ift.tt/fKMetXs I created this because I wanted it and to see how capable Opus 5.5 was. It turned out much better than expected, so I figured I'd share it.
ESPN's Super Bowl LXI broadcast to also stream on Disney+
ESPN announced on Tuesday that its broadcast of Super Bowl LXI on Feb. 14, 2027, will also be streamed on Disney+.
from www.espn.com - TOP https://ift.tt/rP9uoJX
from www.espn.com - TOP https://ift.tt/rP9uoJX
Selasa, 06 Oktober 2026
Rankings for Week 5
Weekly composite ranks for each position in PPR and IDP formats, as well as Eric Karabell's superflex ranks.
from www.espn.com - TOP https://ift.tt/vE2STkW
from www.espn.com - TOP https://ift.tt/vE2STkW
Senin, 05 Oktober 2026
Marauding center backs, do-it-all midfielders: Tactical trends at every position
Elite soccer has changed dramatically in recent years, and the most valued player skill sets have evolved as a result.
from www.espn.com - TOP https://ift.tt/Mf2Ne68
from www.espn.com - TOP https://ift.tt/Mf2Ne68
Minggu, 04 Oktober 2026
'Upgraded to mom and dad': Michael Penix Jr., wife expecting first child
The couple took to Instagram to announce the news, jointly posting a video to the social media site titled "A Short Love Story."
from www.espn.com - TOP https://ift.tt/8W7kiYB
from www.espn.com - TOP https://ift.tt/8W7kiYB
Sabtu, 03 Oktober 2026
Follow live: White Sox, Guardians clash in Game 1 of ALDS
All eight teams are in action Saturday. Which clubs will come out strong?
from www.espn.com - TOP https://ift.tt/P9k13Wp
from www.espn.com - TOP https://ift.tt/P9k13Wp
Your guide to the WNBA semifinals: Keys to Aces-Valkyries and Liberty-Dream
The semifinals open Sunday. How do Las Vegas-Golden State and New York-Atlanta match up? Who are the favorites?
from www.espn.com - TOP https://ift.tt/GVY2KWC
from www.espn.com - TOP https://ift.tt/GVY2KWC
Jumat, 02 Oktober 2026
Sources: Browns' sack leader Graham (knee, ankle) set to miss a few weeks
Browns defensive tackle Mason Graham is expected to miss a few weeks after suffering an MCL sprain and ankle sprain against the Steelers, sources told NFL Network.
from www.espn.com - TOP https://ift.tt/Sy0uqZi
from www.espn.com - TOP https://ift.tt/Sy0uqZi
Romo won't return as NFL analyst, CBS Sports says
Tony Romo will not return as an NFL game analyst for CBS Sports, the network announced, after he was arrested in July on charges of operating a vehicle while intoxicated.
from www.espn.com - TOP https://ift.tt/PAFnbJ0
from www.espn.com - TOP https://ift.tt/PAFnbJ0
Kamis, 01 Oktober 2026
New top story on Hacker News: Show HN: Open-source model routing for coding agents at Astra-level performance
Show HN: Open-source model routing for coding agents at Astra-level performance
6 by adchurch | 0 comments on Hacker News.
A few months ago we started building a model router for coding agents because we thought we could outperform any single model with an ensemble approach. Recently we’ve achieved that milestone and I want to talk about how we did it. First of all, a quick explanation: the Weave Router ( https://ift.tt/HfW7y2I ) plugs into any coding agent (e.g. Claude Code or Codex) and intelligently switches between LLMs. So, for example, Astra handles tricky debugging or complex system design tasks, and Deepseek v4 Flash handles simple frontend updates. What we’re announcing today is our new routing model, which we’re calling Weave Router 2.0. We benchmarked 2.0 against GPT-6 Astra on Terminal Bench 4.0 and SWE Atlas. On both benchmarks, the router had equivalent pass rates. On Terminal Bench, the router hit 52% of Astra’s cost, and completed tasks 2.2x faster. On SWE Atlas, the router cost 54% as much as Astra and ran 2.5x faster. (Full results on our website at https://ift.tt/WNLDiIJ !) It turns out training a model to route effectively - taking into consideration model capabilities, costs, cache awareness, and more - is a really hard problem! I want to talk about three ways we were able to improve so much over the last few months: 1) a new architecture, 2) larger training data set size, and 3) smarter cache-eviction impact calculation. 1) a new architecture. Our initial approach used an RL model without many priors. While RL is still an important part of the story, the cost of fully exploring the space of routing decisions is very high, so we’ve taken some shortcuts that have significantly improved performance. Consider how large the search space for the routing problem is. Take a typical coding agent session, with ~100 agent turns (i.e. 100 LLM API calls). Technically there are 100 chances to select a model. If we assume a roster of ~10 models (of course there are lots more but we can remove any that are Pareto dominated), then there are 10^100 possible paths through that session. We simply cannot explore all of them! So that's why clever tricks to shrink this space are so important. In particular: we trained a hidden Markov model to trace the session state, then a classifier maps the session to one of a few buckets of similar models. Using the HMM allows us to evaluate not just where a session is currently, but how it got there . We've gotten significantly better performance on bucket selection by incorporating that information - we believe this is because two sessions that might look quite similar to a naive classifier are much better distinguished by this HMM approach. Using this HMM + classifier to select a bucket first significantly shrinks the space to explore, by throwing out most models that could not reasonably serve the given session. This rearchitecture was the single biggest performance unlock! 2) larger training data set size (much less technically interesting but still an important part of the story). By using frontier LLMs to help us label a larger and more diverse set of coding agent sessions, we were able to bootstrap the two models discussed in 1) to a better state, while also providing even richer reward signals for RL. 3) smarter cache-eviction impact calculation. One of the hardest parts of routing well (if you care about saving money) is using the model caches intelligently. We built a subsystem that can calculate the expected value of switching models (and thus paying a high one-time cost to fill up a different cache) much more accurately, helping us avoid costly and unnecessary switches in more cases, while still switching when the benefit outweighs the cost. This is where most of our improvement on cost has come from. We still have a lot of room to continue to improve (we won’t rest until we’re consistently beating Astra/Fable, not just tying!) but matching frontier model performance was a huge milestone for our routing model, and in my opinion validates our initial hypothesis that an ensemble of models can do better than any single model ever could. Our router is open source ( https://ift.tt/HfW7y2I ) so anyone can try it out. Or if you prefer you can use our hosted version ( https://ift.tt/WNLDiIJ ).
6 by adchurch | 0 comments on Hacker News.
A few months ago we started building a model router for coding agents because we thought we could outperform any single model with an ensemble approach. Recently we’ve achieved that milestone and I want to talk about how we did it. First of all, a quick explanation: the Weave Router ( https://ift.tt/HfW7y2I ) plugs into any coding agent (e.g. Claude Code or Codex) and intelligently switches between LLMs. So, for example, Astra handles tricky debugging or complex system design tasks, and Deepseek v4 Flash handles simple frontend updates. What we’re announcing today is our new routing model, which we’re calling Weave Router 2.0. We benchmarked 2.0 against GPT-6 Astra on Terminal Bench 4.0 and SWE Atlas. On both benchmarks, the router had equivalent pass rates. On Terminal Bench, the router hit 52% of Astra’s cost, and completed tasks 2.2x faster. On SWE Atlas, the router cost 54% as much as Astra and ran 2.5x faster. (Full results on our website at https://ift.tt/WNLDiIJ !) It turns out training a model to route effectively - taking into consideration model capabilities, costs, cache awareness, and more - is a really hard problem! I want to talk about three ways we were able to improve so much over the last few months: 1) a new architecture, 2) larger training data set size, and 3) smarter cache-eviction impact calculation. 1) a new architecture. Our initial approach used an RL model without many priors. While RL is still an important part of the story, the cost of fully exploring the space of routing decisions is very high, so we’ve taken some shortcuts that have significantly improved performance. Consider how large the search space for the routing problem is. Take a typical coding agent session, with ~100 agent turns (i.e. 100 LLM API calls). Technically there are 100 chances to select a model. If we assume a roster of ~10 models (of course there are lots more but we can remove any that are Pareto dominated), then there are 10^100 possible paths through that session. We simply cannot explore all of them! So that's why clever tricks to shrink this space are so important. In particular: we trained a hidden Markov model to trace the session state, then a classifier maps the session to one of a few buckets of similar models. Using the HMM allows us to evaluate not just where a session is currently, but how it got there . We've gotten significantly better performance on bucket selection by incorporating that information - we believe this is because two sessions that might look quite similar to a naive classifier are much better distinguished by this HMM approach. Using this HMM + classifier to select a bucket first significantly shrinks the space to explore, by throwing out most models that could not reasonably serve the given session. This rearchitecture was the single biggest performance unlock! 2) larger training data set size (much less technically interesting but still an important part of the story). By using frontier LLMs to help us label a larger and more diverse set of coding agent sessions, we were able to bootstrap the two models discussed in 1) to a better state, while also providing even richer reward signals for RL. 3) smarter cache-eviction impact calculation. One of the hardest parts of routing well (if you care about saving money) is using the model caches intelligently. We built a subsystem that can calculate the expected value of switching models (and thus paying a high one-time cost to fill up a different cache) much more accurately, helping us avoid costly and unnecessary switches in more cases, while still switching when the benefit outweighs the cost. This is where most of our improvement on cost has come from. We still have a lot of room to continue to improve (we won’t rest until we’re consistently beating Astra/Fable, not just tying!) but matching frontier model performance was a huge milestone for our routing model, and in my opinion validates our initial hypothesis that an ensemble of models can do better than any single model ever could. Our router is open source ( https://ift.tt/HfW7y2I ) so anyone can try it out. Or if you prefer you can use our hosted version ( https://ift.tt/WNLDiIJ ).
NIT, WBIT cutting down to 24 teams in wake of expanded NCAA tournaments
In the aftermath of the NCAA tournaments expanding to 76 teams starting this season, the NIT and Women's Basketball Invitation Tournament will drop from 32 teams to 24 beginning in 2027, the NCAA announced.
from www.espn.com - TOP https://ift.tt/um5YNfs
from www.espn.com - TOP https://ift.tt/um5YNfs
Langganan:
Postingan (Atom)