Загрузка видео...

Не удалось загрузить видео

На главную

We asked James Wang why Cerebras can run large models so quickly. His answer: Inference speed is mostly a memory problem. Instead of constantly pulling weights from external memory, Cerebras splits the model across multiple wafers and pipelines the layers together. “Inference is all about memory bandwidth.” “You don’t...

MTSlive's profile picture

MTS

212,995 subscribers

27,935 просмотров • 1 месяц назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

SITUATION EXPLAINED: Cerebras raised $5.55 billion in their IPO and closing their first day of trading valued at $66 billion, making it the biggest US tech IPO since Snowflake in 2020. Cerebras makes Wafer-Scale Engine chips built for AI inference. We asked Sarah Fong the main difference between wafer-scale chips and traditional GPUs: - GPUs are great at parallel work (graphics, training) - AI inference is sequential, AKA one token at a time This causes the "memory wall" problem: - Every GPU core needs model weights, KV cache, and activations to do its math - On a GPU, that data lives in off-chip memory (HBM) - Cores constantly load and offload from off-chip memory, which is a huge bottleneck; hardware accounts for ~70% of inference latency Cerebras' chips: -Dinner-plate sized (vs. GPUs which are palm-sized) with tens of thousands of cores -Memory sits directly on top of the cores as distributed SRAM -Weights and KV cache can be accessed at on-chip speeds in the PB/s range, compared with off-chip speeds in the TB/s range achieved by GPUs with HBM.
3:14

SITUATION EXPLAINED: Cerebras raised $5.55 billion in their IPO and closing their first day of trading valued at $66 billion, making it the biggest US tech IPO since Snowflake in 2020. Cerebras makes Wafer-Scale Engine chips built for AI inference. We asked Sarah Fong the main difference between wafer-scale chips and traditional GPUs: - GPUs are great at parallel work (graphics, training) - AI inference is sequential, AKA one token at a time This causes the "memory wall" problem: - Every GPU core needs model weights, KV cache, and activations to do its math - On a GPU, that data lives in off-chip memory (HBM) - Cores constantly load and offload from off-chip memory, which is a huge bottleneck; hardware accounts for ~70% of inference latency Cerebras' chips: -Dinner-plate sized (vs. GPUs which are palm-sized) with tens of thousands of cores -Memory sits directly on top of the cores as distributed SRAM -Weights and KV cache can be accessed at on-chip speeds in the PB/s range, compared with off-chip speeds in the TB/s range achieved by GPUs with HBM.

MTS

44,057 просмотров • 1 месяц назад

Cerebras just IPO’d and the stock already ran up over 100% (Save this). For the entire 70 year history of the semiconductor industry, every company on earth has followed the same process. You take a dinner plate sized silicon wafer, put hundreds of tiny chips onto it, and dice it up like a pizza. Nvidia does it this way, AMD does it this way, Intel has done it this way for six decades and everyone who tried to break that convention failed. Until Cerebras asked the most annoyingly obvious question in the industry’s history, what if you just didn’t cut it? The result is the Wafer Scale Engine, a single chip 56 times larger than Nvidia’s H100 and it fundamentally changes the physics of how AI inference works. The reason this matters is not the size, it’s the bandwidth. Every time an AI model generates a single word, it has to reach into memory, pull weights, multiply them together, and produce a prediction and when you’re running millions of concurrent sessions at once, the bottleneck is not raw processing power but how fast data moves between memory and compute. Nvidia’s H100 moves data at roughly 3 terabytes per second, while Cerebras’ WSE-3 moves data at 21 petabytes per second, roughly 7,000 times faster because memory and compute live on the same enormous piece of silicon and data barely has to travel at all. That gap is exactly why OpenAI went from 150 tokens per second on traditional GPUs to 2,000 tokens per second on Cerebras hardware, and why AWS integrated Cerebras into Bedrock to deliver roughly 5x more inference capacity in the same physical footprint. The macro setup is making the trade even more urgent. South Korea DRAM export prices recently jumped 35%, flash memory surged 47%, and SSD pricing spiked nearly 140% and every single one of those increases hits Nvidia-based infrastructure directly, because the H100 requires 80GB of the most expensive, most contested memory in the AI supply chain. Cerebras’ WSE-3 uses zero external HBM memory, baking 44GB of SRAM directly into the wafer itself which means as memory pricing goes parabolic, every CFO evaluating AI infrastructure is suddenly looking much more seriously at the architecture that sidesteps that cost entirely. The demand is already showing up in the backlog. Cerebras ended 2025 with $24.6 billion in remaining performance obligations for a company doing just over $500 million in annual revenue, that is a number that implies years of contracted growth already sitting on the books. The IPO was 20x oversubscribed, the price range was raised twice before listing, and shares opened 89% above their listing price on a $5.55 billion raise that made it the largest semiconductor IPO in history. The risks are real and worth naming. 86% of 2025 revenue came from two entities with UAE ties, U.S. revenue actually fell 34% to $187 million, and the $20 billion OpenAI contract is conditional, if Cerebras misses delivery milestones, OpenAI can terminate and trigger repayment demands on a $1 billion loan facility. And yet the market is valuing Cerebras at roughly 91x trailing revenue, richer than Nvidia, AMD, and Arm combined. What investors are betting on is not that Cerebras beats Nvidia, it is that the inference supercycle is large enough to support an entirely different architecture optimized for a different workload, and that $24.6 billion in contracted backlog converts to diversified revenue before the market starts asking harder questions. CEO Andrew Feldman said this took a decade of late nights to get right, everyone who tried to copy it failed and given that the entire inference economy is now running through exactly the bottleneck Cerebras was built to eliminate, the market is starting to believe him.
2:04

Cerebras just IPO’d and the stock already ran up over 100% (Save this). For the entire 70 year history of the semiconductor industry, every company on earth has followed the same process. You take a dinner plate sized silicon wafer, put hundreds of tiny chips onto it, and dice it up like a pizza. Nvidia does it this way, AMD does it this way, Intel has done it this way for six decades and everyone who tried to break that convention failed. Until Cerebras asked the most annoyingly obvious question in the industry’s history, what if you just didn’t cut it? The result is the Wafer Scale Engine, a single chip 56 times larger than Nvidia’s H100 and it fundamentally changes the physics of how AI inference works. The reason this matters is not the size, it’s the bandwidth. Every time an AI model generates a single word, it has to reach into memory, pull weights, multiply them together, and produce a prediction and when you’re running millions of concurrent sessions at once, the bottleneck is not raw processing power but how fast data moves between memory and compute. Nvidia’s H100 moves data at roughly 3 terabytes per second, while Cerebras’ WSE-3 moves data at 21 petabytes per second, roughly 7,000 times faster because memory and compute live on the same enormous piece of silicon and data barely has to travel at all. That gap is exactly why OpenAI went from 150 tokens per second on traditional GPUs to 2,000 tokens per second on Cerebras hardware, and why AWS integrated Cerebras into Bedrock to deliver roughly 5x more inference capacity in the same physical footprint. The macro setup is making the trade even more urgent. South Korea DRAM export prices recently jumped 35%, flash memory surged 47%, and SSD pricing spiked nearly 140% and every single one of those increases hits Nvidia-based infrastructure directly, because the H100 requires 80GB of the most expensive, most contested memory in the AI supply chain. Cerebras’ WSE-3 uses zero external HBM memory, baking 44GB of SRAM directly into the wafer itself which means as memory pricing goes parabolic, every CFO evaluating AI infrastructure is suddenly looking much more seriously at the architecture that sidesteps that cost entirely. The demand is already showing up in the backlog. Cerebras ended 2025 with $24.6 billion in remaining performance obligations for a company doing just over $500 million in annual revenue, that is a number that implies years of contracted growth already sitting on the books. The IPO was 20x oversubscribed, the price range was raised twice before listing, and shares opened 89% above their listing price on a $5.55 billion raise that made it the largest semiconductor IPO in history. The risks are real and worth naming. 86% of 2025 revenue came from two entities with UAE ties, U.S. revenue actually fell 34% to $187 million, and the $20 billion OpenAI contract is conditional, if Cerebras misses delivery milestones, OpenAI can terminate and trigger repayment demands on a $1 billion loan facility. And yet the market is valuing Cerebras at roughly 91x trailing revenue, richer than Nvidia, AMD, and Arm combined. What investors are betting on is not that Cerebras beats Nvidia, it is that the inference supercycle is large enough to support an entirely different architecture optimized for a different workload, and that $24.6 billion in contracted backlog converts to diversified revenue before the market starts asking harder questions. CEO Andrew Feldman said this took a decade of late nights to get right, everyone who tried to copy it failed and given that the entire inference economy is now running through exactly the bottleneck Cerebras was built to eliminate, the market is starting to believe him.

Milk Road AI

30,441 просмотров • 2 месяцев назад

Micron is going to $4,000 and once you understand what inference actually is, the number stops sounding crazy (Save this). Dylan Patel just said that by 2030, OpenAI and Anthropic alone will need over 100 gigawatts of compute combined and by 2040, we may not even be measuring AI infrastructure in gigawatts anymore. We may be talking about terawatts. Every single one of those gigawatts needs memory to function. Without it, the compute is worthless. Most people heard that and thought about Nvidia but they should be thinking about Micron. Every AI model generating a response has two phases. The first is prefill, processing your prompt which is compute-heavy and the second is decode generating each word one token at a time and that phase is almost entirely memory-bound, not compute-bound. During decode, the GPU's processing units sit idle more than 95% of the time, waiting for data to arrive from memory. Google confirmed it in a research paper that decode-phase bottlenecks are dominated by memory bandwidth and capacity not raw compute. The GPU is not the bottleneck but the memory feeding the GPU is. This matters because inference is now where all the money lives. Training a model happens once, Inference happens billions of times a day every ChatGPT response, every Claude output, every agentic workflow running in the background and every one of those token streams is a billing event tied directly to memory performance. Adding more GPUs does not fix this because GPUs are already underutilized in inference because they are sitting idle waiting on memory. Adding more memory bandwidth and capacity is what directly reduces token cost, reduces latency, and allows the same cluster to serve dramatically more users simultaneously. Longer context windows compound the problem further, a model running a 1 million token context window requires dramatically more memory per session than a 10,000 token window, and every new model generation pushes context longer. The market treats memory as a downstream beneficiary of Nvidia orders. The correct framework is the opposite, Micron is the upstream constraint on how much value every Nvidia GPU can actually generate at inference scale. Micron guided Q4 to $50 billion in revenue, has HBM4 ramping at twice the pace of the prior generation, and CEO Sanjay Mehrotra has said supply will not catch demand before the end of 2027. At 8x forward earnings on $112 projected FY2027 EPS, Micron is the most undervalued infrastructure company in the entire AI stack. Inference is memory. Memory is Micron and the inference ramp has barely started. Milk Road Pro members are already up massively on this position and we're just getting started. If you want the full breakdown of what we're buying and why, come join us for just a dollar using the link below!
0:39

Micron is going to $4,000 and once you understand what inference actually is, the number stops sounding crazy (Save this). Dylan Patel just said that by 2030, OpenAI and Anthropic alone will need over 100 gigawatts of compute combined and by 2040, we may not even be measuring AI infrastructure in gigawatts anymore. We may be talking about terawatts. Every single one of those gigawatts needs memory to function. Without it, the compute is worthless. Most people heard that and thought about Nvidia but they should be thinking about Micron. Every AI model generating a response has two phases. The first is prefill, processing your prompt which is compute-heavy and the second is decode generating each word one token at a time and that phase is almost entirely memory-bound, not compute-bound. During decode, the GPU's processing units sit idle more than 95% of the time, waiting for data to arrive from memory. Google confirmed it in a research paper that decode-phase bottlenecks are dominated by memory bandwidth and capacity not raw compute. The GPU is not the bottleneck but the memory feeding the GPU is. This matters because inference is now where all the money lives. Training a model happens once, Inference happens billions of times a day every ChatGPT response, every Claude output, every agentic workflow running in the background and every one of those token streams is a billing event tied directly to memory performance. Adding more GPUs does not fix this because GPUs are already underutilized in inference because they are sitting idle waiting on memory. Adding more memory bandwidth and capacity is what directly reduces token cost, reduces latency, and allows the same cluster to serve dramatically more users simultaneously. Longer context windows compound the problem further, a model running a 1 million token context window requires dramatically more memory per session than a 10,000 token window, and every new model generation pushes context longer. The market treats memory as a downstream beneficiary of Nvidia orders. The correct framework is the opposite, Micron is the upstream constraint on how much value every Nvidia GPU can actually generate at inference scale. Micron guided Q4 to $50 billion in revenue, has HBM4 ramping at twice the pace of the prior generation, and CEO Sanjay Mehrotra has said supply will not catch demand before the end of 2027. At 8x forward earnings on $112 projected FY2027 EPS, Micron is the most undervalued infrastructure company in the entire AI stack. Inference is memory. Memory is Micron and the inference ramp has barely started. Milk Road Pro members are already up massively on this position and we're just getting started. If you want the full breakdown of what we're buying and why, come join us for just a dollar using the link below!

Milk Road AI

128,079 просмотров • 15 дней назад

Etched is deploying two new technologies in chip design: low-voltage inference and cluster-scale memory. CEO Gavin Uberti says they'll make their chips much more power-efficient and way, way faster than today's leading GPUs. He breaks it down: "We looked at a lot of early research directions, and we realized the key things that models need are way more compute and way faster memory." "If you think about inference, there are two key parts: prefill and decode. For prefill, it's a compute-bound problem. You need to have more FLOPS, more operations per second on each of your chips." "On our GPU, the bottleneck's actually thermals. You can't really run a GPU at more than around 50% of what it could theoretically do, or it'll melt." "So we're using a new technology today called low-voltage inference to try to solve this problem. You bring the voltage of the chip down dramatically, which allows us to have way, way better efficiency in terms of how much power is drawn per unit of math, and thus fit way way more flops onto the chip..." "For decode, it's all about bandwidth. Not just bandwidth on a chip, but bandwidth across your cluster. That's why we have this technology we call cluster-scale memory. It reduces the amount of time it takes to communicate from one chip to another dramatically." "As a result we can go use all of our HBM, HBM bandwidth, SRAM, SRAM bandwidth, and our scale-up domain as a single coherent pool. And that means if you're a user, you can go get much faster tokens per second, while still keeping your costs low."
3:19

Etched is deploying two new technologies in chip design: low-voltage inference and cluster-scale memory. CEO Gavin Uberti says they'll make their chips much more power-efficient and way, way faster than today's leading GPUs. He breaks it down: "We looked at a lot of early research directions, and we realized the key things that models need are way more compute and way faster memory." "If you think about inference, there are two key parts: prefill and decode. For prefill, it's a compute-bound problem. You need to have more FLOPS, more operations per second on each of your chips." "On our GPU, the bottleneck's actually thermals. You can't really run a GPU at more than around 50% of what it could theoretically do, or it'll melt." "So we're using a new technology today called low-voltage inference to try to solve this problem. You bring the voltage of the chip down dramatically, which allows us to have way, way better efficiency in terms of how much power is drawn per unit of math, and thus fit way way more flops onto the chip..." "For decode, it's all about bandwidth. Not just bandwidth on a chip, but bandwidth across your cluster. That's why we have this technology we call cluster-scale memory. It reduces the amount of time it takes to communicate from one chip to another dramatically." "As a result we can go use all of our HBM, HBM bandwidth, SRAM, SRAM bandwidth, and our scale-up domain as a single coherent pool. And that means if you're a user, you can go get much faster tokens per second, while still keeping your costs low."

TBPN

20,404 просмотров • 15 дней назад

NVIDIA sent us 2 DGX Sparks. For a while we wondered what we would do with them. The memory bandwidth is 273GB/s making it 3x slower than an M3 Ultra (819GB/s) for batch_size=1 inference. But it has 4x more FLOPS (100 TFLOPS compared to 26 TFLOPS). So we thought, what if we could combine the DGX Spark & M3 Ultra, and make use of both the massive compute on the DGX Spark and the massive memory-bandwidth on the M3 Ultra. We came up with a way to split inference across both devices and achieve a speedup of up to 4x for long prompts compared to the M3 Ultra on its own. Full details in the blog post linked below.
1:13

NVIDIA sent us 2 DGX Sparks. For a while we wondered what we would do with them. The memory bandwidth is 273GB/s making it 3x slower than an M3 Ultra (819GB/s) for batch_size=1 inference. But it has 4x more FLOPS (100 TFLOPS compared to 26 TFLOPS). So we thought, what if we could combine the DGX Spark & M3 Ultra, and make use of both the massive compute on the DGX Spark and the massive memory-bandwidth on the M3 Ultra. We came up with a way to split inference across both devices and achieve a speedup of up to 4x for long prompts compared to the M3 Ultra on its own. Full details in the blog post linked below.

Alex Cheema

281,225 просмотров • 9 месяцев назад

The creator of High Bandwidth Memory (HBM) put a number on the AI build that should stop every infra investor cold. A cluster of a million GPUs runs at roughly 10-20% utilization (Save this). Kim Jung-ho spent thirty years building what feeds the GPU, and his claim is that the GPU is barely working. Here is what is actually happening. Every time a model generates output, the data has to be read out of memory, computed, and written back. The read and the write swallow almost the entire cycle. While that data moves, the GPU does nothing. It sits there, fully powered, fully paid for, waiting. By Kim's estimate the memory is doing only about 30 percent of the work it needs to do. The processor idles the rest. So a million installed GPUs run at 10 to 20 percent. You are not compute constrained. You are memory constrained, and the expensive part is standing around. Adding more GPUs does not fix this. It gives you more processors starving for the same data. Here is the part that decides the next decade. Memory can grow. When a cell cannot shrink any further, you stack it into a high-rise, layer on layer. A GPU cannot be stacked. It runs too hot and needs a cooler bolted to its back, so the one move that rescues memory is closed to the processor. The thing that can keep stacking compounds. The thing that cannot plateaus. The marginal dollar in an AI build now buys more by fixing the memory path than by bolting on another idle GPU. Which is why the companies that control memory bandwidth and supply are not suppliers to the AI trade. They are the AI trade.
0:50

The creator of High Bandwidth Memory (HBM) put a number on the AI build that should stop every infra investor cold. A cluster of a million GPUs runs at roughly 10-20% utilization (Save this). Kim Jung-ho spent thirty years building what feeds the GPU, and his claim is that the GPU is barely working. Here is what is actually happening. Every time a model generates output, the data has to be read out of memory, computed, and written back. The read and the write swallow almost the entire cycle. While that data moves, the GPU does nothing. It sits there, fully powered, fully paid for, waiting. By Kim's estimate the memory is doing only about 30 percent of the work it needs to do. The processor idles the rest. So a million installed GPUs run at 10 to 20 percent. You are not compute constrained. You are memory constrained, and the expensive part is standing around. Adding more GPUs does not fix this. It gives you more processors starving for the same data. Here is the part that decides the next decade. Memory can grow. When a cell cannot shrink any further, you stack it into a high-rise, layer on layer. A GPU cannot be stacked. It runs too hot and needs a cooler bolted to its back, so the one move that rescues memory is closed to the processor. The thing that can keep stacking compounds. The thing that cannot plateaus. The marginal dollar in an AI build now buys more by fixing the memory path than by bolting on another idle GPU. Which is why the companies that control memory bandwidth and supply are not suppliers to the AI trade. They are the AI trade.

Fireside Alpha

38,370 просмотров • 16 дней назад

I had to test it myself to believe this unreal inference speed. 3,000 tokens/s for 1 user on standard datacenter GPUs. They leveraged a hidden efficiency gap in how GPUs generate tokens. Kog just achieved 3,000 tokens/s on 8× AMD MI300X GPUs and 2,100 on 8× NVIDIA H200 (FP16, no speculative decoding). Their tech preview is on a 2B model, and they show how their techniques will scale to large frontier MoE models at similar speeds. That's a huge number because normal low-batch GPU decoding for 2B to 8B models is usually closer to 100 to 300 tokens/s per request, so Kog is claiming something like a 10X to 30X jump in the speed one user actually feels. Their trick: they are getting the speed by treating LLM decoding as a memory streaming problem, not mainly a math problem. For 1 user at batch size 1, the GPU is not doing big, efficient matrix-matrix work like in training or large-batch serving; it is repeatedly pulling the model’s active weights from high-bandwidth memory for each new token, so speed depends on how smoothly those weights keep flowing. Normal inference stacks keep breaking that flow. They run many separate GPU programs for different parts of the model, move intermediate results through memory, wait at synchronization points, talk back to the CPU for scheduling or sampling, and then repeat this token after token. Kog’s answer is to co-design 3 things that are usually tuned separately: the runtime, the low-level GPU code, and the model architecture. The biggest engineering move is the monokernel, where the whole decode pass runs as 1 persistent GPU-resident program, including sampling, so the system does not keep stopping for kernel launches, CPU scheduling, and intermediate memory round trips. They also rebuilt synchronization, because their own measurements say grid sync was eating around 35% of token-generation time; instead of making every compute unit wait at a broad barrier, each unit waits only for the exact data it needs. On AMD MI300X, they also map memory access around the chiplet layout, because memory latency changes depending on which die makes the request. Then their Laneformer model uses Delayed Tensor Parallelism, which lets cross-GPU communication happen in the background instead of blocking every layer.
1:08

I had to test it myself to believe this unreal inference speed. 3,000 tokens/s for 1 user on standard datacenter GPUs. They leveraged a hidden efficiency gap in how GPUs generate tokens. Kog just achieved 3,000 tokens/s on 8× AMD MI300X GPUs and 2,100 on 8× NVIDIA H200 (FP16, no speculative decoding). Their tech preview is on a 2B model, and they show how their techniques will scale to large frontier MoE models at similar speeds. That's a huge number because normal low-batch GPU decoding for 2B to 8B models is usually closer to 100 to 300 tokens/s per request, so Kog is claiming something like a 10X to 30X jump in the speed one user actually feels. Their trick: they are getting the speed by treating LLM decoding as a memory streaming problem, not mainly a math problem. For 1 user at batch size 1, the GPU is not doing big, efficient matrix-matrix work like in training or large-batch serving; it is repeatedly pulling the model’s active weights from high-bandwidth memory for each new token, so speed depends on how smoothly those weights keep flowing. Normal inference stacks keep breaking that flow. They run many separate GPU programs for different parts of the model, move intermediate results through memory, wait at synchronization points, talk back to the CPU for scheduling or sampling, and then repeat this token after token. Kog’s answer is to co-design 3 things that are usually tuned separately: the runtime, the low-level GPU code, and the model architecture. The biggest engineering move is the monokernel, where the whole decode pass runs as 1 persistent GPU-resident program, including sampling, so the system does not keep stopping for kernel launches, CPU scheduling, and intermediate memory round trips. They also rebuilt synchronization, because their own measurements say grid sync was eating around 35% of token-generation time; instead of making every compute unit wait at a broad barrier, each unit waits only for the exact data it needs. On AMD MI300X, they also map memory access around the chiplet layout, because memory latency changes depending on which die makes the request. Then their Laneformer model uses Delayed Tensor Parallelism, which lets cross-GPU communication happen in the background instead of blocking every layer.

Rohan Paul

13,148 просмотров • 1 месяц назад

New course: Agent Memory: Building Memory-Aware Agents, built in partnership with Oracle and taught by Richmond Alake and Nacho Martínez. Many agents work well within a single session but their memory resets once the session ends. Consider a research agent working on dozens of papers across multiple days: without memory, it has no way to store and retrieve what it learned across sessions. This short course teaches you to build a memory system that enables agents to persist memory and thereby learn across sessions. You'll design a Memory Manager that handles different memory types, implement semantic tool retrieval that scales without bloating the context, and build write-back pipelines that let your agent autonomously update and refine what it knows over time. Skills you'll gain: - Build persistent memory stores for different agent memory types - Implement a Memory Manager that orchestrates how your agent reads, writes, and retrieves memory - Treat tools as procedural memory and retrieve only relevant ones at inference time using semantic search Join and learn to build agents that remember and improve over time!
1:37

New course: Agent Memory: Building Memory-Aware Agents, built in partnership with Oracle and taught by Richmond Alake and Nacho Martínez. Many agents work well within a single session but their memory resets once the session ends. Consider a research agent working on dozens of papers across multiple days: without memory, it has no way to store and retrieve what it learned across sessions. This short course teaches you to build a memory system that enables agents to persist memory and thereby learn across sessions. You'll design a Memory Manager that handles different memory types, implement semantic tool retrieval that scales without bloating the context, and build write-back pipelines that let your agent autonomously update and refine what it knows over time. Skills you'll gain: - Build persistent memory stores for different agent memory types - Implement a Memory Manager that orchestrates how your agent reads, writes, and retrieves memory - Treat tools as procedural memory and retrieve only relevant ones at inference time using semantic search Join and learn to build agents that remember and improve over time!

Andrew Ng

159,484 просмотров • 3 месяцев назад

most agent memory is one file that grows forever and gets re-read every single turn. it overflows the context, overwrites old facts, loses the thread. that's why your agent feels sharp on day one and lost by week three. Sibyl Memory replaces the pile with a structure: → leaner context, lower token cost. a bloated agent file and flat memory get re-read in full every message. Sibyl Memory keeps a light hot layer and retrieves the rest on demand, so each turn carries only the tokens that matter. → relations are first-class. people link to projects, projects to deals, deals to the decisions behind them. ask about one partner and the agent surfaces the whole connected web around them. that relational context is what lets it help run a company. → one source of truth per fact, person, and project. no duplicates, no silent overwrites. works with hermes, claude code, and codex. beta open.
0:34

most agent memory is one file that grows forever and gets re-read every single turn. it overflows the context, overwrites old facts, loses the thread. that's why your agent feels sharp on day one and lost by week three. Sibyl Memory replaces the pile with a structure: → leaner context, lower token cost. a bloated agent file and flat memory get re-read in full every message. Sibyl Memory keeps a light hot layer and retrieves the rest on demand, so each turn carries only the tokens that matter. → relations are first-class. people link to projects, projects to deals, deals to the decisions behind them. ask about one partner and the agent surfaces the whole connected web around them. that relational context is what lets it help run a company. → one source of truth per fact, person, and project. no duplicates, no silent overwrites. works with hermes, claude code, and codex. beta open.

SIBYL

81,361 просмотров • 1 месяц назад

Palantir CEO Alex Karp on why the model is not the moat. The application layer is: "To make them valuable in an enterprise, like a battlefield, regulated, or manufacturing context, you have to have what's called an application layer." "We have this thing called ontology that now everyone's copying. It takes a large language model and makes it safe and useful and precise." "Safe because it doesn't touch your underlying data." "Safe because it prevents the large language model from caching your data and replicating your business." The tell for any enterprise AI pitch: ask what sits between the model and your data. If the answer is nothing, you are not buying leverage. You are renting a fast way to hand your business to whoever owns the weights. Alex Karp, CEO of Palantir, on CNBC.
0:24

Palantir CEO Alex Karp on why the model is not the moat. The application layer is: "To make them valuable in an enterprise, like a battlefield, regulated, or manufacturing context, you have to have what's called an application layer." "We have this thing called ontology that now everyone's copying. It takes a large language model and makes it safe and useful and precise." "Safe because it doesn't touch your underlying data." "Safe because it prevents the large language model from caching your data and replicating your business." The tell for any enterprise AI pitch: ask what sits between the model and your data. If the answer is nothing, you are not buying leverage. You are renting a fast way to hand your business to whoever owns the weights. Alex Karp, CEO of Palantir, on CNBC.

Karl Mehta

31,447 просмотров • 10 дней назад

We think #GenAI will transform virtually every customer experience. That’s why we’re doubling down on all three layers of what’s called the GenAI “stack.” What do we mean by stack? Simple. Three layers, bottom to top. The “bottom layer” means the infrastructure. Think chips, servers, networking – the software, hardware, and computing power in our data centers used to train, build, and deploy AI models and applications. The “middle layer” is the GenAI sandbox – all the tools, Large Language Models (LLMs), and Foundation Models (FMs) we’ve built into Amazon Bedrock. This lets customers pick the #AI models that best fit their needs to easily and securely customize GenAI applications specific to their business. The “top layer” means the popular agents and assistants that people actually interact with – now including Amazon Q. These are built from the models in the middle layer – in Amazon Bedrock. While other apps might help you write a quick email, or edit an image, Amazon Q’s meant for taking care of business. It’s been engineered to be the most capable GenAI-powered assistant for speeding up software dev and leveraging companies’ internal data. Are you as bullish on the future of GenAI as we are? Listen to our CEO Andy Jassy talk about our investment across the stack.
1:19

We think #GenAI will transform virtually every customer experience. That’s why we’re doubling down on all three layers of what’s called the GenAI “stack.” What do we mean by stack? Simple. Three layers, bottom to top. The “bottom layer” means the infrastructure. Think chips, servers, networking – the software, hardware, and computing power in our data centers used to train, build, and deploy AI models and applications. The “middle layer” is the GenAI sandbox – all the tools, Large Language Models (LLMs), and Foundation Models (FMs) we’ve built into Amazon Bedrock. This lets customers pick the #AI models that best fit their needs to easily and securely customize GenAI applications specific to their business. The “top layer” means the popular agents and assistants that people actually interact with – now including Amazon Q. These are built from the models in the middle layer – in Amazon Bedrock. While other apps might help you write a quick email, or edit an image, Amazon Q’s meant for taking care of business. It’s been engineered to be the most capable GenAI-powered assistant for speeding up software dev and leveraging companies’ internal data. Are you as bullish on the future of GenAI as we are? Listen to our CEO Andy Jassy talk about our investment across the stack.

Amazon

47,421 просмотров • 2 лет назад

$CBRS is a very unique company because it keeps a huge amount of compute and memory very close together which helps it deliver extremely fast inference and makes the architecture well positioned for agentic AI workloads where fast tokens matter more. OpenAI validated the tec but the stock is getting hit because we still don't have enough visibility into how Cerebras turns that technical advantage into a broad, diversified and high-margin business. Customer concentration remains a massive risk especially with so much revenue tied to its own investor and the company still has to solve the physical infrastructure side of the business across power, cooling, data center capacity and deployment timing. If $NVDA is trading at a below-market multiple while still growing faster than Cerebras then why buy the Nvidia alternative instead of just owning Nvidia?
0:59

$CBRS is a very unique company because it keeps a huge amount of compute and memory very close together which helps it deliver extremely fast inference and makes the architecture well positioned for agentic AI workloads where fast tokens matter more. OpenAI validated the tec but the stock is getting hit because we still don't have enough visibility into how Cerebras turns that technical advantage into a broad, diversified and high-margin business. Customer concentration remains a massive risk especially with so much revenue tied to its own investor and the company still has to solve the physical infrastructure side of the business across power, cooling, data center capacity and deployment timing. If $NVDA is trading at a below-market multiple while still growing faster than Cerebras then why buy the Nvidia alternative instead of just owning Nvidia?

Shay Boloor

190,537 просмотров • 20 дней назад

$NBIS cofounder Roman Chernin describes how their recent acquisitions of Eigen AI and Clarifai were all about speed, incredible talent, and acceleration: "The philosophy is very simple. We need to build so many things, and we need to move so fast, that we're always looking for people who can accelerate us. It should be exceptional talent, and/or something that has a great adoption." "Our two recent acquisitions [were] two teams that work on inference optimization. A big part of our business is how efficiently we convert GPUs into tokens. And these two teams — Eigen AI and Clarifai — one is focused on model optimization, the engine of inference. How you run specific models and all the techniques around spec decoding, quantization, and so on." "And the other is system optimization. All the routing, KV caching, and orchestration across the big cluster of compute and so on." "We have a very strong internal team working on inference. But we felt that we needed to move faster, bring more capabilities. Because the market is so fast."
2:32

$NBIS cofounder Roman Chernin describes how their recent acquisitions of Eigen AI and Clarifai were all about speed, incredible talent, and acceleration: "The philosophy is very simple. We need to build so many things, and we need to move so fast, that we're always looking for people who can accelerate us. It should be exceptional talent, and/or something that has a great adoption." "Our two recent acquisitions [were] two teams that work on inference optimization. A big part of our business is how efficiently we convert GPUs into tokens. And these two teams — Eigen AI and Clarifai — one is focused on model optimization, the engine of inference. How you run specific models and all the techniques around spec decoding, quantization, and so on." "And the other is system optimization. All the routing, KV caching, and orchestration across the big cluster of compute and so on." "We have a very strong internal team working on inference. But we felt that we needed to move faster, bring more capabilities. Because the market is so fast."

TBPN

30,486 просмотров • 2 месяцев назад

Fireworks AI is now available first-party in Microsoft Foundry, as part of a multi-year partnership, starting today in private preview. Enterprise AI has a speed problem. Most infrastructure can't keep up with how fast open models are moving. New architectures, fine-tuned weights, models that aren't on anyone's approved list yet. The usual answer was "manage your own serving infrastructure." That's a real cost. Fireworks runs the inference. Azure owns the governance. Your weights stay yours. Beyond custom weights, teams can also access the latest open models on day zero. No waiting for the marketplace to catch up. Serverless inference for prototyping, and a path to production that doesn't require re-architecting anything. Fireworks handles the serving layer. Azure handles the rest. Your boss handles the congratulations. Want in →
0:23

Fireworks AI is now available first-party in Microsoft Foundry, as part of a multi-year partnership, starting today in private preview. Enterprise AI has a speed problem. Most infrastructure can't keep up with how fast open models are moving. New architectures, fine-tuned weights, models that aren't on anyone's approved list yet. The usual answer was "manage your own serving infrastructure." That's a real cost. Fireworks runs the inference. Azure owns the governance. Your weights stay yours. Beyond custom weights, teams can also access the latest open models on day zero. No waiting for the marketplace to catch up. Serverless inference for prototyping, and a path to production that doesn't require re-architecting anything. Fireworks handles the serving layer. Azure handles the rest. Your boss handles the congratulations. Want in →

Fireworks AI

51,657 просмотров • 4 месяцев назад

Autonomous AI agents can’t afford a memory problem. After OpenClaw, insane numbers of autonomous agents are coming online, but all of them have a memory layer that's breaking. I have been playing around with SureThing these days. It is an autonomous AI agent designed with a native memory layer. → it learns your patterns → integrates with 1000+ apps → fully autonomous They actually benchmarked memory depth. SureThing: 84.8, ChatGPT: 40, OpenClaw: 60. The gap speaks for itself. SureThing is invitation-only. The crazy part is that you have to talk to SureThing agent itself to get the invite code. The onboarding is the product demo. You feel the memory and judgment before you even have an account. Try it here: Partnered with SureThing for this post. Thanks to the team for the support.
2:06

Autonomous AI agents can’t afford a memory problem. After OpenClaw, insane numbers of autonomous agents are coming online, but all of them have a memory layer that's breaking. I have been playing around with SureThing these days. It is an autonomous AI agent designed with a native memory layer. → it learns your patterns → integrates with 1000+ apps → fully autonomous They actually benchmarked memory depth. SureThing: 84.8, ChatGPT: 40, OpenClaw: 60. The gap speaks for itself. SureThing is invitation-only. The crazy part is that you have to talk to SureThing agent itself to get the invite code. The onboarding is the product demo. You feel the memory and judgment before you even have an account. Try it here: Partnered with SureThing for this post. Thanks to the team for the support.

Pratham

80,634 просмотров • 4 месяцев назад

A tricky LLM interview question: You're serving a reasoning model on vLLM, and it keeps running out of GPU memory on long traces. So you add KV cache compression and evict 90% of the cached tokens. VRAM usage stays as is and GPU still runs out of memory. Why? (answer below) Evicting 90% of the KV cache can free almost none of the memory it was using. This sounds counterintuitive, but it follows directly from how production servers store the cache today. The KV cache grows with every token a model generates. Each token appends its key and value vectors across every layer, and nothing is freed while generation continues. This is the dominant memory cost for reasoning models. If a 32K-token CoT caches ~32K tokens of KV vectors, a Qwen3-32B with 4-bit weights will run out-of-memory around 24K tokens on a 24GB GPU. One obvious solution is to keep the important tokens and drop the rest, since attention is sparse enough to allow it. But this does not solve the memory problem yet. The reason is paged attention, which is the memory manager behind vLLM and most production servers. Under the hood, it splits GPU memory into fixed physical blocks, each one holds the KV for about 16 tokens. This block returns to the allocator only when every slot inside it is empty. Since the eviction logic selects tokens by importance, and such tokens are scattered across blocks... ...so despite eviction, almost every block is left with at least some survivor tokens. For instance, if the logic evicts 14k of 16k tokens across 1,000 blocks, most likely every block will still have a token. This means the allocator frees almost nothing. Placing the new tokens into those freed slots is not ideal because it breaks the cache's layout. Say token 16,001 arrives, and it's placed in the slot the 40th token used to hold. The cache now reads position 38, then 16,001, then 41, so the cache is no longer in token order. Attention can still compute the right answer from that, but only if every slot now carries a separate note recording which position it actually holds. This introduces another bookkeeping cost that an in-order layout inherently avoids. So the cache is logically 90% smaller and still physically the same size. Many compression results miss this because they measure on pre-allocated contiguous tensors rather than a paged server. There's another problem. Eviction methods pick which tokens to keep by looking at the attention scores themselves (as expected). But fast attention kernels used in production, like FlashAttention, never save those scores. They compute attention in small pieces and throw the full score grid away as they go, which is also why they're fast. So the exact signal eviction methods need isn't available in memory. The workaround is to fall back to eager attention and build the full matrix, which gives up the speed FlashAttention was there to provide. NVIDIA published a method called TriAttention to solve both these problems. It never needs attention scores. Instead, it scores tokens from the geometry of the model's key and query vectors before RoPE is applied, where those vectors sit in stable clusters. For the memory problem, it runs a compaction pass every 128 decoded tokens. The surviving tokens slide forward to close the holes eviction creates, so whole blocks empty out and return to the allocator while the cache stays in token order. On long reasoning traces, the approach matches full-attention accuracy while decoding 2.5x faster and using 10.7x less KV memory. KV cache compression is a big infrastructure problem. The number that decides whether it works is the count of freed blocks, not the count of evicted tokens. You can find the NVIDIA write-up here: I wrote a first-principles breakdown of how the KV cache works. It walks through why the model stores keys and values at all, why the cache grows with every token, and a comparison of LLM generation speed with and without KV caching. Read it below.
0:23

A tricky LLM interview question: You're serving a reasoning model on vLLM, and it keeps running out of GPU memory on long traces. So you add KV cache compression and evict 90% of the cached tokens. VRAM usage stays as is and GPU still runs out of memory. Why? (answer below) Evicting 90% of the KV cache can free almost none of the memory it was using. This sounds counterintuitive, but it follows directly from how production servers store the cache today. The KV cache grows with every token a model generates. Each token appends its key and value vectors across every layer, and nothing is freed while generation continues. This is the dominant memory cost for reasoning models. If a 32K-token CoT caches ~32K tokens of KV vectors, a Qwen3-32B with 4-bit weights will run out-of-memory around 24K tokens on a 24GB GPU. One obvious solution is to keep the important tokens and drop the rest, since attention is sparse enough to allow it. But this does not solve the memory problem yet. The reason is paged attention, which is the memory manager behind vLLM and most production servers. Under the hood, it splits GPU memory into fixed physical blocks, each one holds the KV for about 16 tokens. This block returns to the allocator only when every slot inside it is empty. Since the eviction logic selects tokens by importance, and such tokens are scattered across blocks... ...so despite eviction, almost every block is left with at least some survivor tokens. For instance, if the logic evicts 14k of 16k tokens across 1,000 blocks, most likely every block will still have a token. This means the allocator frees almost nothing. Placing the new tokens into those freed slots is not ideal because it breaks the cache's layout. Say token 16,001 arrives, and it's placed in the slot the 40th token used to hold. The cache now reads position 38, then 16,001, then 41, so the cache is no longer in token order. Attention can still compute the right answer from that, but only if every slot now carries a separate note recording which position it actually holds. This introduces another bookkeeping cost that an in-order layout inherently avoids. So the cache is logically 90% smaller and still physically the same size. Many compression results miss this because they measure on pre-allocated contiguous tensors rather than a paged server. There's another problem. Eviction methods pick which tokens to keep by looking at the attention scores themselves (as expected). But fast attention kernels used in production, like FlashAttention, never save those scores. They compute attention in small pieces and throw the full score grid away as they go, which is also why they're fast. So the exact signal eviction methods need isn't available in memory. The workaround is to fall back to eager attention and build the full matrix, which gives up the speed FlashAttention was there to provide. NVIDIA published a method called TriAttention to solve both these problems. It never needs attention scores. Instead, it scores tokens from the geometry of the model's key and query vectors before RoPE is applied, where those vectors sit in stable clusters. For the memory problem, it runs a compaction pass every 128 decoded tokens. The surviving tokens slide forward to close the holes eviction creates, so whole blocks empty out and return to the allocator while the cache stays in token order. On long reasoning traces, the approach matches full-attention accuracy while decoding 2.5x faster and using 10.7x less KV memory. KV cache compression is a big infrastructure problem. The number that decides whether it works is the count of freed blocks, not the count of evicted tokens. You can find the NVIDIA write-up here: I wrote a first-principles breakdown of how the KV cache works. It walks through why the model stores keys and values at all, why the cache grows with every token, and a comparison of LLM generation speed with and without KV caching. Read it below.

Avi Chawla

267,206 просмотров • 18 дней назад

Yesterday, I joined TBPN to talk about what it really means to give AI a memory. And, why that’s becoming the next big moat. We discussed: - Why today’s AI apps are stateless - How memory is quickly becoming the real moat for AI systems - How Mem0 is decoupling memory from models - Why users should be able to own and carry their AI memories across apps - And, how 200 cold emails changed my life🙂 Appreciate John Coogan, Jordi Hays & the TBPN team for having me and for such a thoughtful discussion around the future of AI infrastructure.
7:08

Yesterday, I joined TBPN to talk about what it really means to give AI a memory. And, why that’s becoming the next big moat. We discussed: - Why today’s AI apps are stateless - How memory is quickly becoming the real moat for AI systems - How Mem0 is decoupling memory from models - Why users should be able to own and carry their AI memories across apps - And, how 200 cold emails changed my life🙂 Appreciate John Coogan, Jordi Hays & the TBPN team for having me and for such a thoughtful discussion around the future of AI infrastructure.

Taranjeet

14,913 просмотров • 8 месяцев назад

This is big. The Retrieval Plugin allows ChatGPT to have a memory! The model can now remember information from conversations and store it in the retrieval plugin for later use. This feature is a must if you want to develop GPT-based tools.
0:38

This is big. The Retrieval Plugin allows ChatGPT to have a memory! The model can now remember information from conversations and store it in the retrieval plugin for later use. This feature is a must if you want to develop GPT-based tools.

Lior Alexander

227,586 просмотров • 3 лет назад

We analyzed Anthropic’s memory architecture and built something better: a persistent, human-inspectable, and token-efficient memory layer that scales with your projects. Today, it’s OPEN-SOURCE. ByteRover CLI gives agents (like OpenClaw, Claude Code, and Hermes) persistent, structured memory. Built on the exact architecture that became the #1 memory system for OpenClaw (30,000+ downloads in a week), it lets developers curate project knowledge into a file-based hierarchy. This guarantees highly accurate, lightning-fast retrieval, even with lightweight models. 👉Highly Accurate: >92% retrieval accuracy across long-running sessions - the highest proven production accuracy on the market. ⚡Fast: ~1.6s average retrieval time. 💰Economic: Maintains >90% accuracy even with lightweight models, saving 50-70% on token costs. ☁ Portable: Runs locally by default, with cloud-sync to share memory across agents and teammates.
1:19

We analyzed Anthropic’s memory architecture and built something better: a persistent, human-inspectable, and token-efficient memory layer that scales with your projects. Today, it’s OPEN-SOURCE. ByteRover CLI gives agents (like OpenClaw, Claude Code, and Hermes) persistent, structured memory. Built on the exact architecture that became the #1 memory system for OpenClaw (30,000+ downloads in a week), it lets developers curate project knowledge into a file-based hierarchy. This guarantees highly accurate, lightning-fast retrieval, even with lightweight models. 👉Highly Accurate: >92% retrieval accuracy across long-running sessions - the highest proven production accuracy on the market. ⚡Fast: ~1.6s average retrieval time. 💰Economic: Maintains >90% accuracy even with lightweight models, saving 50-70% on token costs. ☁ Portable: Runs locally by default, with cloud-sync to share memory across agents and teammates.

andy nguyen

28,980 просмотров • 3 месяцев назад

Memory for OpenClaw is now Native! Our first OpenClaw Memory Skill was a massive success: 30k+ downloads in a week and 500k+ organic impressions overnight for launch post. But we knew memory needed to be native. On March 21, OpenClaw merged PR #50848, allowing us to go beyond the skill layer and integrate directly into the agent’s context assembly flow. We try to make OpenClaw a truly 24/7 employee capable of complex workflows. The technical setup isn’t the hardest part but the real challenge is giving it a "brain" that remembers exact project details, past decisions, and team changes over time. The Native Memory Plugin is now live on NPM & ClawHub. Here is what it brings to your OpenClaw agents: 👉 Native Integration: Automatically manages a Three-Layer Memory architecture (Context Tree, Workspace Memory, Daily Memory). 👉 Git-like Stateful Memory: Organizes memory into a semantic hierarchy of human-readable, diffable Markdown files. You always get updated knowledge and can actually see and fix what your agent learns. 👉 Top Market Accuracy: Achieves an industry-leading 92.2% retrieval accuracy (LoCoMo & LongMemEval benchmarks), maintaining 90% accuracy even with cheap, lightweight models. 👉 Local-first & Portable: Local-by-default, fully portable for multi-agent teams. 👉 Super Easy Setup
1:46

Memory for OpenClaw is now Native! Our first OpenClaw Memory Skill was a massive success: 30k+ downloads in a week and 500k+ organic impressions overnight for launch post. But we knew memory needed to be native. On March 21, OpenClaw merged PR #50848, allowing us to go beyond the skill layer and integrate directly into the agent’s context assembly flow. We try to make OpenClaw a truly 24/7 employee capable of complex workflows. The technical setup isn’t the hardest part but the real challenge is giving it a "brain" that remembers exact project details, past decisions, and team changes over time. The Native Memory Plugin is now live on NPM & ClawHub. Here is what it brings to your OpenClaw agents: 👉 Native Integration: Automatically manages a Three-Layer Memory architecture (Context Tree, Workspace Memory, Daily Memory). 👉 Git-like Stateful Memory: Organizes memory into a semantic hierarchy of human-readable, diffable Markdown files. You always get updated knowledge and can actually see and fix what your agent learns. 👉 Top Market Accuracy: Achieves an industry-leading 92.2% retrieval accuracy (LoCoMo & LongMemEval benchmarks), maintaining 90% accuracy even with cheap, lightweight models. 👉 Local-first & Portable: Local-by-default, fully portable for multi-agent teams. 👉 Super Easy Setup

andy nguyen

169,136 просмотров • 3 месяцев назад