Loading video...

Video Failed to Load

Go Home

SOMEONE BENCHMARKED 10 MODELS ON STACKING BLOCKS IN A PHYSICS SIM its basically jenga in reverse every block you place could topple the whole tower, and going higher always means more risk > opus 5 hit 11.07m using only 15 blocks, then ended its turn to lock in the...

om_patel5's profile picture

Om Patel

26,869 subscribers

16,166 views • 5 days ago •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

🚨 CRYPTO: BITCOIN NETWORK HITS RARE TWO-BLOCK REORGANIZATION Not an attack. Not a bug. Just Bitcoin doing what Bitcoin does. The network briefly split into two competing chains today at block height 941,880. AntPool mined one version of block 941,881, ViaBTC extended it to 941,882. Simultaneously, Foundry USA mined its own competing versions of both blocks. The tie was broken when Foundry kept winning. Seven consecutive blocks (941,879 to 941,885). AntPool and ViaBTC's blocks became orphans. Single-block reorgs happen periodically. Two-block reorgs are much rarer, meaning the tie persisted for an entire extra block cycle. Still well within normal consensus behavior. No $BTC or funds lost. No double spends. Just proof-of-work settling a race the way Satoshi designed it.
0:14

🚨 CRYPTO: BITCOIN NETWORK HITS RARE TWO-BLOCK REORGANIZATION Not an attack. Not a bug. Just Bitcoin doing what Bitcoin does. The network briefly split into two competing chains today at block height 941,880. AntPool mined one version of block 941,881, ViaBTC extended it to 941,882. Simultaneously, Foundry USA mined its own competing versions of both blocks. The tie was broken when Foundry kept winning. Seven consecutive blocks (941,879 to 941,885). AntPool and ViaBTC's blocks became orphans. Single-block reorgs happen periodically. Two-block reorgs are much rarer, meaning the tie persisted for an entire extra block cycle. Still well within normal consensus behavior. No $BTC or funds lost. No double spends. Just proof-of-work settling a race the way Satoshi designed it.

BSCN

16,507 views • 4 months ago

Which local models can actually handle tool calling? I built a framework to find out. 15 scenarios. 12 tools. Mocked responses. Temperature 0. No cherry-picking. Tested every Qwen3.5 size from 0.8B to 397B, and since some of you asked after the distillation tests: yes, I included Jackrong's Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled too. Only two models went all green: the 27B dense and the distilled 27B. The 397B? Failed two tests. The 122B? Failed one. The 35B? Failed two. The timed-out results — mostly on the smaller models, are cases where the model got stuck in a loop, repeating the same tool call until it hit the 30-second limit. The test that exposed the most models: "Search for Iceland's population, then calculate 2% of it." Simple, but 35B, 122B, and 397B all used a rounded number from memory instead of the actual search result. They didn't trust their own tool output. Small models hallucinate data. Big models ignore data. The 27B just threaded it through.
0:11

Which local models can actually handle tool calling? I built a framework to find out. 15 scenarios. 12 tools. Mocked responses. Temperature 0. No cherry-picking. Tested every Qwen3.5 size from 0.8B to 397B, and since some of you asked after the distillation tests: yes, I included Jackrong's Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled too. Only two models went all green: the 27B dense and the distilled 27B. The 397B? Failed two tests. The 122B? Failed one. The 35B? Failed two. The timed-out results — mostly on the smaller models, are cases where the model got stuck in a loop, repeating the same tool call until it hit the 30-second limit. The test that exposed the most models: "Search for Iceland's population, then calculate 2% of it." Simple, but 35B, 122B, and 397B all used a rounded number from memory instead of the actual search result. They didn't trust their own tool output. Small models hallucinate data. Big models ignore data. The 27B just threaded it through.

stevibe

431,569 views • 4 months ago

Kimi K3 beats GPT 5.6 Sol on both speed and in cost in backend bug fixing ! Planted 4 defects in the same Python order service: a wrong-total calculation, a broken access-control check (IDOR), a crash on invalid input, and an inventory oversell. Both models got identical code and the same prompt, with an automated test suite verifying every fix over the live API. Both went 4/4, but by the time Kimi K3 finished all 4, GPT 5.6 had only fixed 2. Kimi wrapped in under half the time, at half the cost. Kimi K3: 1min 34s · $0.1 · 4/4 GPT 5.6 Sol: 4min 6s · $0.2 · 4/4
0:13

Kimi K3 beats GPT 5.6 Sol on both speed and in cost in backend bug fixing ! Planted 4 defects in the same Python order service: a wrong-total calculation, a broken access-control check (IDOR), a crash on invalid input, and an inventory oversell. Both models got identical code and the same prompt, with an automated test suite verifying every fix over the live API. Both went 4/4, but by the time Kimi K3 finished all 4, GPT 5.6 had only fixed 2. Kimi wrapped in under half the time, at half the cost. Kimi K3: 1min 34s · $0.1 · 4/4 GPT 5.6 Sol: 4min 6s · $0.2 · 4/4

GMI Cloud

24,606 views • 22 days ago

⚡️Kaspa 11 #11 - BlockDAG Kaspa's BlockDAG data structure enables its unique GHOSTDAG consensus algorithm to solve the orphan blocks problem, allowing for high frequency block generation and high scalability. Learn all about this feature and others via the link in the thread. The 10BPS 𝆒 Crescendo update is set to be hard-forked into mainnet at DAA Score 110,165,000, projected to occur on May 5, 2025, at approximately 15:00 UTC. ~ 3 Days
0:12

⚡️Kaspa 11 #11 - BlockDAG Kaspa's BlockDAG data structure enables its unique GHOSTDAG consensus algorithm to solve the orphan blocks problem, allowing for high frequency block generation and high scalability. Learn all about this feature and others via the link in the thread. The 10BPS 𝆒 Crescendo update is set to be hard-forked into mainnet at DAA Score 110,165,000, projected to occur on May 5, 2025, at approximately 15:00 UTC. ~ 3 Days

Kaspa

32,642 views • 1 year ago

1-bit Kimi K3 performs at Opus 5 level on 3D physics! We ran our Atomic Chat quant of Kimi K3 locally on 4x B200 against three cloud models and gave them all the same task, to build a giant anvil drop test as a single HTML file with real physics Outputs: K3 1bit (local): 15.8K tokens, $0 API cost Kimi K3 (API): 15.3K tokens, $0.30 API cost Opus 5: 22.8K tokens, $0.77 API cost GPT 5.6: 14.5K tokens, $0.72 API cost All four got the physics right. But only Kimi made a working winch. The drum turns and the chain drags the flat car off the pad. Opus 5 drew the most detail, road markings and sparks on the hit. And you can run a model at this level on your own box now. That still feels insane to us
0:15

1-bit Kimi K3 performs at Opus 5 level on 3D physics! We ran our Atomic Chat quant of Kimi K3 locally on 4x B200 against three cloud models and gave them all the same task, to build a giant anvil drop test as a single HTML file with real physics Outputs: K3 1bit (local): 15.8K tokens, $0 API cost Kimi K3 (API): 15.3K tokens, $0.30 API cost Opus 5: 22.8K tokens, $0.77 API cost GPT 5.6: 14.5K tokens, $0.72 API cost All four got the physics right. But only Kimi made a working winch. The drum turns and the chain drags the flat car off the pad. Opus 5 drew the most detail, road markings and sparks on the hit. And you can run a model at this level on your own box now. That still feels insane to us

atomic.chat

52,841 views • 11 days ago

April 30 • 12:00pm ET Art Blocks + OpenSea “Gift of time” began during my residency in Marfa, Texas, as part of the Art Blocks and OpenSea artist residency program, where a distinct shift in the experience of time became central to the work. In the desert, I felt time move differently. It stretched, slowed, and became something I noticed. After a few days, the rhythm changed. Moments felt longer, attention sharpened, and I became increasingly aware of each moment as it passed. This work comes from that condition. Time is not treated only as a theme, but as a system embedded in the structure of the piece. Different ways of measuring time, such as mechanical cycles, calendars, and lunar phases, are translated into rules that continuously transform the work. The piece does not represent time. It runs on it. Its movement is tied to blockchain time. Even when unseen, it continues to rotate and evolve. When loaded, it synchronizes with the present moment, but it does not begin when it is viewed, and it does not stop when it disappears from the screen. During the residency, I spent hours thinking, sketching, and making connections. Those connections are also visible. Elastic lines, like rubber bands, link elements across the piece, representing how memories connect, how one thought leads to another, and how everything builds over time. These same connections introduce moments where the system attempts to pull itself back, as if trying to regain control. But it never fully resets. It is not a loop. The movement continues, drifting forward, never returning to a fixed state. Visually, the work reveals its own construction. Lines, paths, and rotations expose an internal logic, like looking inside a mechanism. The drawing language recalls diagrams, technical sketches, or the interior of a mechanical watch. It is a system in motion, always active. “Gift of Time” exists because I was given time by Art Blocks, OpenSea, and above all my family. It is my way of saying thank you. It is both a reflection on time and a product of it. April 30 @ 12:00pm ET on Art blocks & OpenSea 1 / 1 / 365 • 0.02 Eth Art Blocks, OpenSea
0:18

April 30 • 12:00pm ET Art Blocks + OpenSea “Gift of time” began during my residency in Marfa, Texas, as part of the Art Blocks and OpenSea artist residency program, where a distinct shift in the experience of time became central to the work. In the desert, I felt time move differently. It stretched, slowed, and became something I noticed. After a few days, the rhythm changed. Moments felt longer, attention sharpened, and I became increasingly aware of each moment as it passed. This work comes from that condition. Time is not treated only as a theme, but as a system embedded in the structure of the piece. Different ways of measuring time, such as mechanical cycles, calendars, and lunar phases, are translated into rules that continuously transform the work. The piece does not represent time. It runs on it. Its movement is tied to blockchain time. Even when unseen, it continues to rotate and evolve. When loaded, it synchronizes with the present moment, but it does not begin when it is viewed, and it does not stop when it disappears from the screen. During the residency, I spent hours thinking, sketching, and making connections. Those connections are also visible. Elastic lines, like rubber bands, link elements across the piece, representing how memories connect, how one thought leads to another, and how everything builds over time. These same connections introduce moments where the system attempts to pull itself back, as if trying to regain control. But it never fully resets. It is not a loop. The movement continues, drifting forward, never returning to a fixed state. Visually, the work reveals its own construction. Lines, paths, and rotations expose an internal logic, like looking inside a mechanism. The drawing language recalls diagrams, technical sketches, or the interior of a mechanical watch. It is a system in motion, always active. “Gift of Time” exists because I was given time by Art Blocks, OpenSea, and above all my family. It is my way of saying thank you. It is both a reflection on time and a product of it. April 30 @ 12:00pm ET on Art blocks & OpenSea 1 / 1 / 365 • 0.02 Eth Art Blocks, OpenSea

Manuel Lariño ☔️

21,901 views • 3 months ago

A guy fed 4.2M expired US patents to Claude and found 6 products nobody manufactures anymore. p.s first one is already in production Every expired patent becomes public. All 4.2M of them are written in legal language nobody actually reads. He asked Claude to process the entire database and score each patent by its attractiveness. All the research, all the problems, all the solutions already documented, someone already paid for that. Just pick the best one and start cooking. Here's how it works: > Python scraper pulls expired patents from USPTO > converts docs to clean Markdown > scores each one 1-10 on viability > applies a filter: score 7+ only > only 1 in 80 selects It found 6 products in 3 weeks and the first one is already generating a profit. how many of those 4.2M are still waiting for you?
0:11

A guy fed 4.2M expired US patents to Claude and found 6 products nobody manufactures anymore. p.s first one is already in production Every expired patent becomes public. All 4.2M of them are written in legal language nobody actually reads. He asked Claude to process the entire database and score each patent by its attractiveness. All the research, all the problems, all the solutions already documented, someone already paid for that. Just pick the best one and start cooking. Here's how it works: > Python scraper pulls expired patents from USPTO > converts docs to clean Markdown > scores each one 1-10 on viability > applies a filter: score 7+ only > only 1 in 80 selects It found 6 products in 3 weeks and the first one is already generating a profit. how many of those 4.2M are still waiting for you?

Spivach

11,704 views • 3 months ago

Pi0 vs. ACT with BBox conditioning 🟦 Not many know you can push ACT to *almost-Pi0* generalisation by conditioning on bounding boxes (BBoxes). How is the training data collected? • Generate BBoxes for all pick-and-place objects in the scene.(I used Gemini) • Pick-and-place targets are selected randomly. • Add the BBox coordinates to the robot’s state. • Overlay the BBoxes in the visualisation so you know what to grab and where to drop. During inference: • Generate BBoxes for every object again. • Click the object you want to pick and its target spot; those BBoxes get added to the robot state. • Let the robot do the work for you 😃 Setup: - Trained ACT for 100k steps and fine-tuned Pi0 for only 20k. - Training data is 60 episodes and had *only* LEGO bricks. - Using single front camera (Laptop in this case) Got the idea from xun in LeRobot discord. Here’s ACT vs Pi0 on a toy car that isn’t in the dataset. 1/3
0:12

Pi0 vs. ACT with BBox conditioning 🟦 Not many know you can push ACT to *almost-Pi0* generalisation by conditioning on bounding boxes (BBoxes). How is the training data collected? • Generate BBoxes for all pick-and-place objects in the scene.(I used Gemini) • Pick-and-place targets are selected randomly. • Add the BBox coordinates to the robot’s state. • Overlay the BBoxes in the visualisation so you know what to grab and where to drop. During inference: • Generate BBoxes for every object again. • Click the object you want to pick and its target spot; those BBoxes get added to the robot state. • Let the robot do the work for you 😃 Setup: - Trained ACT for 100k steps and fine-tuned Pi0 for only 20k. - Training data is 60 episodes and had *only* LEGO bricks. - Using single front camera (Laptop in this case) Got the idea from xun in LeRobot discord. Here’s ACT vs Pi0 on a toy car that isn’t in the dataset. 1/3

Shreyas Gite

34,863 views • 1 year ago

The Khumbu Glacier has revealed another silent tragedy, a body emerging from ice that once kept its secrets buried for years, pictured this season 2026. What was frozen in place is now being exposed as temperatures rise across the Himalaya, turning Everest into one of the clearest frontlines of climate change. This season, the warning signs continue higher up in the Khumbu Icefall, where massive unstable serac blocks have delayed route opening and increased danger for climbers and Sherpas. On Everest, the past is resurfacing, and the future is becoming more uncertain.
0:15

The Khumbu Glacier has revealed another silent tragedy, a body emerging from ice that once kept its secrets buried for years, pictured this season 2026. What was frozen in place is now being exposed as temperatures rise across the Himalaya, turning Everest into one of the clearest frontlines of climate change. This season, the warning signs continue higher up in the Khumbu Icefall, where massive unstable serac blocks have delayed route opening and increased danger for climbers and Sherpas. On Everest, the past is resurfacing, and the future is becoming more uncertain.

Everest Today

106,755 views • 3 months ago

Max Recorded TPS (1 Block) is now on Chainspect! 🚀 So far, we've measured Max Recorded TPS only for a period of 100 blocks. You take the total transactions over the time and divide them by cumulative block time to get at estimate of what TPS a blockchain can churn out consistently during peak workload periods. Now, we're adding the Max Recorded TPS per block. That's the highest-ever burst TPS a chain has handled on record — a measure of its peak performance under high stress. Check out how your fav chain has fared:
0:11

Max Recorded TPS (1 Block) is now on Chainspect! 🚀 So far, we've measured Max Recorded TPS only for a period of 100 blocks. You take the total transactions over the time and divide them by cumulative block time to get at estimate of what TPS a blockchain can churn out consistently during peak workload periods. Now, we're adding the Max Recorded TPS per block. That's the highest-ever burst TPS a chain has handled on record — a measure of its peak performance under high stress. Check out how your fav chain has fared:

Chainspect

49,252 views • 1 year ago

OpenCode Go is now wired into Codex!! The pricing is insane. $10 gets you 10,000 DeepSeek requests every 5 hours (no weekly limits). I converted that into DeepSeek API dollars because I thought I was reading it wrong, and the same 5 hours of usage would run somewhere between $10 and $30 depending on how big your context gets. So one afternoon of use already covers the whole sub. It comes with Kimi K3 as well. Both sit in the picker next to my other models now. Next time we hit a limit in the middle of a loop, we can grab it and keep going.
0:17

OpenCode Go is now wired into Codex!! The pricing is insane. $10 gets you 10,000 DeepSeek requests every 5 hours (no weekly limits). I converted that into DeepSeek API dollars because I thought I was reading it wrong, and the same 5 hours of usage would run somewhere between $10 and $30 depending on how big your context gets. So one afternoon of use already covers the whole sub. It comes with Kimi K3 as well. Both sit in the picker next to my other models now. Next time we hit a limit in the middle of a loop, we can grab it and keep going.

Ziwen

422,162 views • 7 days ago

Just dropped: Horizon. Not just a theme. A new foundation for building online stores on Shopify. One system. Ten new themes. All live today. Horizon incorporates theme blocks - 30+ drag-and-drop components you can remix across your store. Video. Product recs. Rich text. Custom content. Want holographic effects on images as you hover over them? Just say so and AI will code something just for you. We’ve also made some of your most requested updates to the theme editor: • Copy + paste blocks anywhere • Drag + drop across pages • Preview blocks before publishing This is total design freedom. The fastest way to go from idea to store. Horizon is the new starting line for great brands. Let’s build.
0:14

Just dropped: Horizon. Not just a theme. A new foundation for building online stores on Shopify. One system. Ten new themes. All live today. Horizon incorporates theme blocks - 30+ drag-and-drop components you can remix across your store. Video. Product recs. Rich text. Custom content. Want holographic effects on images as you hover over them? Just say so and AI will code something just for you. We’ve also made some of your most requested updates to the theme editor: • Copy + paste blocks anywhere • Drag + drop across pages • Preview blocks before publishing This is total design freedom. The fastest way to go from idea to store. Horizon is the new starting line for great brands. Let’s build.

Harley Finkelstein

31,696 views • 1 year ago

Situation on Aleppo [9/1/2026]: [11:00 AM] During the early hours of the morning, the STG continued to advance and secure areas previously evacuated by Asayish forces: in the north, trenches were taken west of the Youth Housing Complex, as well as in the Barakah Quarries, where another series of trenches and new blocks were taken in the Resafa district, as well as the last blocks of the Al Layramoun Industrial district to the west. In the south, the New Syriac district was completely taken and the area was expanded beyond the Assyrian Quarter district after entering the cemetery area and capturing the Al-Jalaa Sports Club. Meanwhile, from the east, the assault on the Sheikh Maqsud district began and the first blocks of the Awared neighborhood were entered. All these advances took place prior to the start of a ceasefire brokered by the US and France, in which the Damascus authorities issued a new ultimatum for the surrender of the enclave in exchange for the evacuation of the inhabitants and Asayish soldiers. Despite the arrival of a convoy of buses in the morning, the Asayish rejected the evacuation and declared their commitment to continue defending the inhabitants of the enclave, given the danger of a new massacre similar to those on the Syrian coast or in Suwayda. Meanwhile, in the east of the Euphrates, a general mobilization has begun in solidarity with the besieged troops, in anticipation of the fighting spreading to the rest of the territory under SDF control. Map: [ ]
0:06

Situation on Aleppo [9/1/2026]: [11:00 AM] During the early hours of the morning, the STG continued to advance and secure areas previously evacuated by Asayish forces: in the north, trenches were taken west of the Youth Housing Complex, as well as in the Barakah Quarries, where another series of trenches and new blocks were taken in the Resafa district, as well as the last blocks of the Al Layramoun Industrial district to the west. In the south, the New Syriac district was completely taken and the area was expanded beyond the Assyrian Quarter district after entering the cemetery area and capturing the Al-Jalaa Sports Club. Meanwhile, from the east, the assault on the Sheikh Maqsud district began and the first blocks of the Awared neighborhood were entered. All these advances took place prior to the start of a ceasefire brokered by the US and France, in which the Damascus authorities issued a new ultimatum for the surrender of the enclave in exchange for the evacuation of the inhabitants and Asayish soldiers. Despite the arrival of a convoy of buses in the morning, the Asayish rejected the evacuation and declared their commitment to continue defending the inhabitants of the enclave, given the danger of a new massacre similar to those on the Syrian coast or in Suwayda. Meanwhile, in the east of the Euphrates, a general mobilization has begun in solidarity with the besieged troops, in anticipation of the fighting spreading to the rest of the territory under SDF control. Map: [ ]

@Suriyak

21,749 views • 7 months ago

Codex can now run Deepseek-v4- flash! There's a catch though. Deepseek's official setup switches your entire codex over to them, so your GPT models stop showing up at all. This is exactly what Codex Router is for. It adds models to the list instead of replacing them, so sol, grok, kimi and deepseek all sit in the same picker and i just grab whichever one suits the job. Deepseek v4-flash is $0.28 per million output tokens. opus 4.8 is $25. same picker, 89x apart. Links in the comment. setup's in the video 👇
0:15

Codex can now run Deepseek-v4- flash! There's a catch though. Deepseek's official setup switches your entire codex over to them, so your GPT models stop showing up at all. This is exactly what Codex Router is for. It adds models to the list instead of replacing them, so sol, grok, kimi and deepseek all sit in the same picker and i just grab whichever one suits the job. Deepseek v4-flash is $0.28 per million output tokens. opus 4.8 is $25. same picker, 89x apart. Links in the comment. setup's in the video 👇

Ziwen

145,018 views • 12 days ago

Here's a quick overview of what we can or could expect from #Kaspa in 2025. - At least 10 blocks per second with a PoW BlockDAG - Listings on major exchanges such as #Binance, #OKX or #Coinbase - A stablecoin on the $KAS BlockDAG - DAGKnight as a new consensus mechanism - Possibly smart contracts - And probably more partnerships in the energy sector and other interesting areas All in all, very bullish and, in my opinion, the most exciting year for Kaspa so far. Especially because a lot of people are mainly looking at the price, I am sure that these people will also be very happy soon. What have I forgotten and what do you think will have the biggest impact on KAS?
0:18

Here's a quick overview of what we can or could expect from #Kaspa in 2025. - At least 10 blocks per second with a PoW BlockDAG - Listings on major exchanges such as #Binance, #OKX or #Coinbase - A stablecoin on the $KAS BlockDAG - DAGKnight as a new consensus mechanism - Possibly smart contracts - And probably more partnerships in the energy sector and other interesting areas All in all, very bullish and, in my opinion, the most exciting year for Kaspa so far. Especially because a lot of people are mainly looking at the price, I am sure that these people will also be very happy soon. What have I forgotten and what do you think will have the biggest impact on KAS?

KASPA Enthusiast 𐤊

19,368 views • 1 year ago

SOMEONE GOT TIRED OF PAYING HIGGSFIELD AI'S SUBSCRIPTION SO HE REBUILT THE WHOLE THING AND OPEN-SOURCED IT 200+ models. text-to-image, image-to-image, text-to-video, image-to-video all in one interface you configure a virtual camera in the Cinema Studio. pick the body, the lens, the focal length, the aperture and it writes the optimized cinematic prompt for you. completely in the background you never touch the camera keywords. you just set up the shot like a real cinematographer would Kling v3, Sora 2, Veo 3, Flux Dev, Midjourney v7, GPT-4o, Seedream 5.0, Runway Gen-3 all in there self-hosted. MIT licensed. runs on your machine. your data stays local the only thing you pay for is the model API calls themselves someone built this so you never have to pay Higgsfield AI again
0:18

SOMEONE GOT TIRED OF PAYING HIGGSFIELD AI'S SUBSCRIPTION SO HE REBUILT THE WHOLE THING AND OPEN-SOURCED IT 200+ models. text-to-image, image-to-image, text-to-video, image-to-video all in one interface you configure a virtual camera in the Cinema Studio. pick the body, the lens, the focal length, the aperture and it writes the optimized cinematic prompt for you. completely in the background you never touch the camera keywords. you just set up the shot like a real cinematographer would Kling v3, Sora 2, Veo 3, Flux Dev, Midjourney v7, GPT-4o, Seedream 5.0, Runway Gen-3 all in there self-hosted. MIT licensed. runs on your machine. your data stays local the only thing you pay for is the model API calls themselves someone built this so you never have to pay Higgsfield AI again

Rimsha Bhardwaj

101,198 views • 4 months ago

Richard Jefferson on The Block being the greatest human play of all time.👇#NBA “Then I played with Duncan and then I got f***** LeBron [James] and you’re just like bro…his intelligence was insane. The greatest play that I’ve ever saw — it’s the block. It’s the block because that’s game 7, he’s played every minute, it’s the last possible game, it’s the last possible minute, and he does an athletic feat of sprinting back, split second late that’s a goal tending, and it’s like, this dude led the series in points rebounds assists steals blocks — in EVERY category. At the last minute when he should be gassed. So the block is the most amazing, the most singular play, but when I look at plays of any human being that have ever been done and the moment they were done, I can’t think of a bigger play.”
0:10

Richard Jefferson on The Block being the greatest human play of all time.👇#NBA “Then I played with Duncan and then I got f***** LeBron [James] and you’re just like bro…his intelligence was insane. The greatest play that I’ve ever saw — it’s the block. It’s the block because that’s game 7, he’s played every minute, it’s the last possible game, it’s the last possible minute, and he does an athletic feat of sprinting back, split second late that’s a goal tending, and it’s like, this dude led the series in points rebounds assists steals blocks — in EVERY category. At the last minute when he should be gassed. So the block is the most amazing, the most singular play, but when I look at plays of any human being that have ever been done and the moment they were done, I can’t think of a bigger play.”

🎗NBA•Fan🎗

4,436,001 views • 3 years ago

PSA: $NAM Airdrop 🪂 - only a few hours left ➡️Claim closes on: 12th of Jan 2024, at 9am ➡️ The _ONLY_ place to claim ➡️Allocations have been adjusted recently - check if you have claimed to see if your number went up! ➡️US is sadly geoblocked 🔸Eligibility and How to Claim reminder ➡️ $OSMO snapshot: mainnet block 12146298 ➡️ $ATOM snapshots: block 17660694 (the last blocks before 1st of November 2023). To be eligible you need to have at least $100 USD staked ➡️In $OSMO: 280 $OSMO ➡️In $ATOM: 12.786667 $ATOM Beyond staking every account gets an Activity Score. The TLDR; of this scoring method is that it measures the engagement of the account on a decentralized network. The specific framework used in the RPGF Drop is an adaptation from Trusta AI ' Media Score, and was first introduced by Celestia on the Genesis Drop. Based on the mechanism, the score can be a minimum of 1.5 and a maximum of 15. The Activity Score is a points system that takes into account activity such as transactions, ibc transfers, the value of transactions, months active, etc. (the scoring details are linked in the announcement). The maximum points you can get on the Activity Score is 15. To determine the tier, the amount staked at the time is multiplied by the Activity Score. Then all eligible accounts are ranked by highest value to lowest and grouped into tiers: top 10%, 50-90%, and 0-50% ➡️Source: MEDIA SCORE as the Infrastructure for On-chain User Value Assessment: 🔸How to claim $NAM airdrop ➡️Get Namada Wallet Extension ➡️Go to ➡️Connect Wallet $ATOM $OSMO $STARS ➡️Connect $NAM Extension ➡️CLAIM $NAM ➡️Repeat for any additional chains / wallets 🔸Looking for an easy and super clear "how to claim video"? @L1am_Crypto & Stakecito Labs to the rescue 🫡 🔸$NAM airdrop beraindex overview: 🔸Stake $ATOM with our sponsor, Enigma , and don't miss out on future Airdrops 🥩 $ATOM 🫡
0:10

PSA: $NAM Airdrop 🪂 - only a few hours left ➡️Claim closes on: 12th of Jan 2024, at 9am ➡️ The _ONLY_ place to claim ➡️Allocations have been adjusted recently - check if you have claimed to see if your number went up! ➡️US is sadly geoblocked 🔸Eligibility and How to Claim reminder ➡️ $OSMO snapshot: mainnet block 12146298 ➡️ $ATOM snapshots: block 17660694 (the last blocks before 1st of November 2023). To be eligible you need to have at least $100 USD staked ➡️In $OSMO: 280 $OSMO ➡️In $ATOM: 12.786667 $ATOM Beyond staking every account gets an Activity Score. The TLDR; of this scoring method is that it measures the engagement of the account on a decentralized network. The specific framework used in the RPGF Drop is an adaptation from Trusta AI ' Media Score, and was first introduced by Celestia on the Genesis Drop. Based on the mechanism, the score can be a minimum of 1.5 and a maximum of 15. The Activity Score is a points system that takes into account activity such as transactions, ibc transfers, the value of transactions, months active, etc. (the scoring details are linked in the announcement). The maximum points you can get on the Activity Score is 15. To determine the tier, the amount staked at the time is multiplied by the Activity Score. Then all eligible accounts are ranked by highest value to lowest and grouped into tiers: top 10%, 50-90%, and 0-50% ➡️Source: MEDIA SCORE as the Infrastructure for On-chain User Value Assessment: 🔸How to claim $NAM airdrop ➡️Get Namada Wallet Extension ➡️Go to ➡️Connect Wallet $ATOM $OSMO $STARS ➡️Connect $NAM Extension ➡️CLAIM $NAM ➡️Repeat for any additional chains / wallets 🔸Looking for an easy and super clear "how to claim video"? @L1am_Crypto & Stakecito Labs to the rescue 🫡 🔸$NAM airdrop beraindex overview: 🔸Stake $ATOM with our sponsor, Enigma , and don't miss out on future Airdrops 🥩 $ATOM 🫡

Grey Ledger

60,319 views • 2 years ago

Stateless History Node is almost like a regular Ethereum node, but it doesn't store state and it doesn't have EVM execution. It's used only for syncing events and thus - is faster and gives you FREE INDEXING. You don't have to pay 6 figures for RPC anymore! Just spin up a Stateless History Node, plug rindexer or Ponder there, and enjoy free (AND FAST!!) indexing! This node is syncing >1000 blocks per second at my local pc (less than 6hrs for the whole Ethereum), and it should use less than 200GB - which means you can host it on a MacMini, Hetzner or whatever. You can futhermore filter that by using block ranges or bloom filters, etc - I haven't developed this yet. What you see is a proof of concept. It works via native devp2p 'eth' protocol, but with EIP4444 and The Prune we would have to also support era1 archives and Portal Network. But so far it works - there are plenty of peers serving historical receipts, and they serve them FAST! If you run Stateless History Node you can also serve the blocks and receipts - so that could help to preserve archival data too. For now there is no data validation yet (and even no data storage - that's a very early PoC), but we can verify validity of chain by simultaneously running a lightweight CL node (or not lightweight if you're extremely paranoid). And then support verifying the hashes of receipts and blocks with their parents, maintaining full integrity and zero trust. It's also written in rust, btw. So, I guess, at least for Ethereum Mainnet the era of RPC's pumping moneybags is over - there's finally a local, trustless and free indexing alternative available. Too sad this won't work for Optimism / Base , cause despite introducing P2P after Bedrock - they haven't enabled receipts transfer in the protocol (or at least I couldn't find one). Arbitrum is even sadder - I don't believe there is a P2P layer at all - you just have to run your own node, hold state and execute blocks to get events. There is hope - Paradigm recently released Ress - stateless execution, but it requires nodes to support Witness preparation & exchange - but this could work for L2s - cause the main blocker for local RPCs rn is huge state (VPS with TB storage cost a lot), and the second blocker is EVM forks makes it hard to hold a node - it needs to be maintained, upgraded, etc. Ress at least solves the state part. But anyways, I will try to continue working on this and release some MVP version with RPC endpoint and data storage soon - follow the updates!
0:16

Stateless History Node is almost like a regular Ethereum node, but it doesn't store state and it doesn't have EVM execution. It's used only for syncing events and thus - is faster and gives you FREE INDEXING. You don't have to pay 6 figures for RPC anymore! Just spin up a Stateless History Node, plug rindexer or Ponder there, and enjoy free (AND FAST!!) indexing! This node is syncing >1000 blocks per second at my local pc (less than 6hrs for the whole Ethereum), and it should use less than 200GB - which means you can host it on a MacMini, Hetzner or whatever. You can futhermore filter that by using block ranges or bloom filters, etc - I haven't developed this yet. What you see is a proof of concept. It works via native devp2p 'eth' protocol, but with EIP4444 and The Prune we would have to also support era1 archives and Portal Network. But so far it works - there are plenty of peers serving historical receipts, and they serve them FAST! If you run Stateless History Node you can also serve the blocks and receipts - so that could help to preserve archival data too. For now there is no data validation yet (and even no data storage - that's a very early PoC), but we can verify validity of chain by simultaneously running a lightweight CL node (or not lightweight if you're extremely paranoid). And then support verifying the hashes of receipts and blocks with their parents, maintaining full integrity and zero trust. It's also written in rust, btw. So, I guess, at least for Ethereum Mainnet the era of RPC's pumping moneybags is over - there's finally a local, trustless and free indexing alternative available. Too sad this won't work for Optimism / Base , cause despite introducing P2P after Bedrock - they haven't enabled receipts transfer in the protocol (or at least I couldn't find one). Arbitrum is even sadder - I don't believe there is a P2P layer at all - you just have to run your own node, hold state and execute blocks to get events. There is hope - Paradigm recently released Ress - stateless execution, but it requires nodes to support Witness preparation & exchange - but this could work for L2s - cause the main blocker for local RPCs rn is huge state (VPS with TB storage cost a lot), and the second blocker is EVM forks makes it hard to hold a node - it needs to be maintained, upgraded, etc. Ress at least solves the state part. But anyways, I will try to continue working on this and release some MVP version with RPC endpoint and data storage soon - follow the updates!

Convergence Boy

28,877 views • 6 months ago

Day 12/90 of Inference Engineering What is chunked prefill within vLLM? In continuation of yesterday's post on the high level architecture of vLLM, I want to dive deeper into vLLM core engine starting with the mechanics of chunked prefill. In this post, I will closely follow the original blog on the anatomy of vLLM. To start, let's define chunked prefill. It's a runtime inference optimization technique that splits a long input request so that it doesn’t monopolize the whole GPU. Keep in mind this is all within the context of vLLM. And since vLLM is an inference engine that's meant to serve a model to multiple concurrent users, having a GPU that’s fully monopolized on a single user's request means other users' requests would be in queue waiting to be processed. It isn’t too good to have the whole GPU occupied on a single request when the GPU is meant to be shared! So the key idea behind chunked prefill is to break the long request into smaller chunks, so that each chunk along with other users' requests gets processed and written into the KV cache together. Suppose we split up the long request into chunks and each chunk has 8 tokens. Now each memory block can hold 4 tokens. Therefore, 8 tokens can fit into 2 blocks of memory. After the first forward pass, 2 blocks are occupied, and after the second forward pass, 4 blocks of memory are occupied and so forth. Each forward pass handles a small chunk of the long request so that there's room in the same pass to keep serving other users' requests. Here's a small animation that I made today to fully visualize the idea behind chunked prefill when learning this topic~
0:18

Day 12/90 of Inference Engineering What is chunked prefill within vLLM? In continuation of yesterday's post on the high level architecture of vLLM, I want to dive deeper into vLLM core engine starting with the mechanics of chunked prefill. In this post, I will closely follow the original blog on the anatomy of vLLM. To start, let's define chunked prefill. It's a runtime inference optimization technique that splits a long input request so that it doesn’t monopolize the whole GPU. Keep in mind this is all within the context of vLLM. And since vLLM is an inference engine that's meant to serve a model to multiple concurrent users, having a GPU that’s fully monopolized on a single user's request means other users' requests would be in queue waiting to be processed. It isn’t too good to have the whole GPU occupied on a single request when the GPU is meant to be shared! So the key idea behind chunked prefill is to break the long request into smaller chunks, so that each chunk along with other users' requests gets processed and written into the KV cache together. Suppose we split up the long request into chunks and each chunk has 8 tokens. Now each memory block can hold 4 tokens. Therefore, 8 tokens can fit into 2 blocks of memory. After the first forward pass, 2 blocks are occupied, and after the second forward pass, 4 blocks of memory are occupied and so forth. Each forward pass handles a small chunk of the long request so that there's room in the same pass to keep serving other users' requests. Here's a small animation that I made today to fully visualize the idea behind chunked prefill when learning this topic~

max fu

29,197 views • 25 days ago