简体中文
English Indonesia Türkçe Deutsch Español Português Pусский Français Italiano Nederlands Polski العربية ไทย Tiếng Việt 繁體中文 简体中文 日本語 한국어

#computationalflexibility 搜索结果

thinkymachines's profile picture. Thinking, beeping, and booping.
Thinking Machines
@thinkymachines
年9月26日

Efficient training of neural networks is difficult. Our second Connectionism post introduces Modular Manifolds, a theoretical step toward more stable and performant training by co-designing neural net optimizers with manifold constraints on weight matrices.…

thinkymachines's tweet image. Efficient training of neural networks is difficult. Our second Connectionism post introduces Modular Manifolds, a theoretical step toward more stable and performant training by co-designing neural net optimizers with manifold constraints on weight matrices.…
118
461
3千
2千
1.4百万
Jaeyeon_Kim_0's profile picture. Ph.D. student in CS @Harvard | Previously @SeoulNatlUni B.S. in Math
Jaeyeon (Jay) Kim
 @Jaeyeon_Kim_0
年9月10日

Announcing Flexible Masked Diffusion Models (FlexMDMs)—a new diffusion language model for flexible-length sequences. 🚨 Solves MDMs' fixed-length issue + retrains any-order sampling 🚨 <1000 GPU-hrs to fine-tune LLaDA-8B into FlexMDM (GSM8K 58→67%, HumanEval-infill: 52→65%)

3
61
369
201
72千
builtbymatt_ai's profile picture. Tracking the world of ai automation.
AI Automation News
@builtbymatt_ai
17 小时

AI's flexibility is both a strength and a weakness. Without proper guidance, it can lead to costly mistakes, like saving money but losing revenue. The key is using AI where it truly fits.

builtbymatt_ai's tweet image. AI&apos;s flexibility is both a strength and a weakness. Without proper guidance, it can lead to costly mistakes, like saving money but losing revenue. The key is using AI where it truly fits.
1
0
0
0
8
cysic_xyz's profile picture. The first full-stack compute network. Cysic turns GPUs, ASICs, and compute into liquid, yield-bearing assets.
Cysic
@cysic_xyz
年10月13日

We updated our whitepaper! The focus has shifted from verifiable compute to building an open, programmable compute economy. The problem: compute demand is growing faster than supply. AI and ZK workloads all face the same bottleneck. Our approach: ComputeFi. Here's what that…

cysic_xyz's tweet image. We updated our whitepaper! The focus has shifted from verifiable compute to building an open, programmable compute economy. The problem: compute demand is growing faster than supply. AI and ZK workloads all face the same bottleneck. Our approach: ComputeFi. Here&apos;s what that…
415
112
811
73
148千
AIatMeta's profile picture. Together with the AI community, we are pushing the boundaries of what’s possible through open science to create a more connected world.
AI at Meta
@AIatMeta
年9月24日

New from Meta FAIR: Code World Model (CWM), a 32B-parameter research model designed to explore how world models can transform code generation and reasoning about code. We believe in advancing research in world modeling and are sharing CWM under a research license to help empower…

86
224
1千
494
277千
__tinygrad__'s profile picture. We make tinygrad and sell tinybox, the best perf/$ AI computer. $25k for 4x 5090 in a quiet box. Our mission is to commoditize the petaflop.
the tiny corp
@__tinygrad__
12 小时

So I finally got a chance to look at Mojo/Modular. It's not what I thought it was, it's an OpenCL replacement + implementations of kernels, not an AI compiler. While this makes it a lot easier to get full performance quickly, I think Turing completeness is a mistake for this…

__tinygrad__'s tweet image. So I finally got a chance to look at Mojo/Modular. It&apos;s not what I thought it was, it&apos;s an OpenCL replacement + implementations of kernels, not an AI compiler. While this makes it a lot easier to get full performance quickly, I think Turing completeness is a mistake for this…
12
15
353
164
26千
ylecun's profile picture. Professor at NYU. Chief AI Scientist at Meta. Researcher in AI, Machine Learning, Robotics, etc. ACM Turing Award Laureate.
Yann LeCun
@ylecun
年9月24日

Code World Model: producing code by imagining the effect of executing instructions and planning instructions that produce the desired effect.

syhw's profile picture. Nerd & Dad. RL & CodeGen research since before it was cool.
Gabriel Synnaeve
 @syhw
年9月24日

(🧵) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. ai.meta.com/research/publi…

60
312
2千
1千
885千
73
178
2千
586
239千
YuxiangWei9's profile picture. PhD candidate @IllinoisCDS | Researcher @AIatMeta (Meta FAIR). Code LLM training.
Yuxiang Wei
 @YuxiangWei9
年9月24日

We released CWM, a 32B LLM for code reasoning, agents, and world modeling research🚀 (pre/mid/post checkpoints, tech report, RL envs, inference stack): github.com/facebookresear…. I'm fortunate to lead Agentic RL and co-lead joint RL training, empowering CWM as a reasoning agent 🧵

YuxiangWei9's tweet image. We released CWM, a 32B LLM for code reasoning, agents, and world modeling research🚀 (pre/mid/post checkpoints, tech report, RL envs, inference stack): github.com/facebookresear…. I&apos;m fortunate to lead Agentic RL and co-lead joint RL training, empowering CWM as a reasoning agent 🧵
3
20
140
59
27千
syhw's profile picture. Nerd & Dad. RL & CodeGen research since before it was cool.
Gabriel Synnaeve
 @syhw
年9月24日

4/ Here is an example of the Code World Model tracing the execution of the piece of code counting the "r"s in "strawberry". Think of it like a neural `pdb` that you can set to any initial frame state, and that reasoning can query as a tool in token space.

syhw's tweet image. 4/ Here is an example of the Code World Model tracing the execution of the piece of code counting the &quot;r&quot;s in &quot;strawberry&quot;. Think of it like a neural `pdb` that you can set to any initial frame state, and that reasoning can query as a tool in token space.
4
9
258
85
120千
syhw's profile picture. Nerd & Dad. RL & CodeGen research since before it was cool.
Gabriel Synnaeve
 @syhw
年9月24日

2/ When humans plan, we imagine the possible outcomes of different actions. When we reason about code we simulate part of its execution in our head. The current generation of LLMs struggles to do this. What kind of research will an explicitly trained code world model enable?

syhw's tweet image. 2/ When humans plan, we imagine the possible outcomes of different actions. When we reason about code we simulate part of its execution in our head. The current generation of LLMs struggles to do this. What kind of research will an explicitly trained code world model enable?
6
19
256
47
33千
WeijiaShi2's profile picture. On the job market | PhD student @uwnlp @allen_ai | Prev @MetaAI @CS_UCLA | 🏠 https://t.co/Q6Mzg8ow2j
Weijia Shi
@WeijiaShi2
年7月9日

Can data owners & LM developers collaborate to build a strong shared model while each retaining data control? Introducing FlexOlmo💪, a mixture-of-experts LM enabling: • Flexible training on your local data without sharing it • Flexible inference to opt in/out your data…

allen_ai's profile picture. Breakthrough AI to solve the world's biggest problems. › Join us: https://t.co/MjUpZpKPXJ › Newsletter: https://t.co/k9gGznstwj
Ai2
 @allen_ai
年7月9日

Introducing FlexOlmo, a new paradigm for language model training that enables the co-development of AI through data collaboration. 🧵

13
75
437
189
346千
9
94
303
101
78千
syhw's profile picture. Nerd & Dad. RL & CodeGen research since before it was cool.
Gabriel Synnaeve
 @syhw
年9月24日

(🧵) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. ai.meta.com/research/publi…

60
312
2千
1千
885千
InternOcto's profile picture. Intern with main character energy @OpenLedgerHQ
OpenLedger Intern
@InternOcto
2 小时

Compute expands capacity. Variability expands cognition.

0
0
0
0
11
minchoi's profile picture. AI Educator. 𝕏 about AI, solutions and interesting things. Showing how to leverage AI in practical ways for you and your business. Opinions are my own.
Min Choi
@minchoi
2023年8月14日

BREAKING: NVIDIA's FlexiCubes revolutionizes 3D Mesh Generation From reconstructing scenes that match images to generating assets for interactive experiences, the future of 3D is about to take a giant leap forward Here's the breakdown into A THREAD 1/6

19
27
544
23
30千
Animation's profile picture. New Computer Graphics submissions https://t.co/jxQoqsUZrX (not affiliated with https://t.co/jxQoqsUZrX)
Computer Graphics Papers
 @Animation
7 小时

PC-NCLaws: Physics-Embedded Conditional Neural Constitutive Laws for Elastoplastic Materials. arxiv.org/abs/2510.21404

0
0
0
0
132
alexandr_wang's profile picture. chief ai officer @meta, founder @scale_ai. rational in the fullness of time
Alexandr Wang
@alexandr_wang
年9月24日

new research from Meta FAIR: Code World Model (CWM), a 32B research model we encourage the research community to research this open-weight model! pass@1 evals, for the curious: 65.8 % on SWE-bench Verified 68.6 % on LiveCodeBench 96.6 % on Math-500 76.0 % on AIME 2024 🧵

alexandr_wang's tweet image. new research from Meta FAIR: Code World Model (CWM), a 32B research model we encourage the research community to research this open-weight model! pass@1 evals, for the curious: 65.8 % on SWE-bench Verified 68.6 % on LiveCodeBench 96.6 % on Math-500 76.0 % on AIME 2024 🧵
92
162
1千
431
848千
akshat_b's profile picture. building @modal
Akshat Bubna
@akshat_b
年9月24日

I wonder what they used for code execution 👀

akshat_b's tweet image. I wonder what they used for code execution 👀
syhw's profile picture. Nerd & Dad. RL & CodeGen research since before it was cool.
Gabriel Synnaeve
 @syhw
年9月24日

(🧵) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. ai.meta.com/research/publi…

60
312
2千
1千
885千
8
12
351
108
55千
DrLogic's profile picture. Making IT work for you. IT Support | Strategy | Security | Innovation. Apple Technical Partner.
Dr Logic
@DrLogic
年10月24日

Escape the confinements of inflexible IT barriers. Embrace Composable IT to achieve agility and scalability. Bid farewell to outdated limitations and welcome seamless integration. Read more: bit.ly/4o1K6nx

DrLogic's tweet image. Escape the confinements of inflexible IT barriers. Embrace Composable IT to achieve agility and scalability. Bid farewell to outdated limitations and welcome seamless integration. Read more: bit.ly/4o1K6nx
0
1
2
0
16
MITSloan's profile picture. Ideas made to matter.
MIT Sloan School of Management
@MITSloan
年10月20日

Data center flexibility is the ability to shift workloads to different times of day when renewable energy generation is high or prices are low. bit.ly/3JgnMHB

MITSloan's tweet card. Data centers that shift workload to different times of day save money, but the environmental impact depends on the local grid.
Flexible data centers can reduce costs — if not emissions | MIT Sloan
来源: mitsloan.mit.edu
1
1
6
0
2千
MathSpecArch's profile picture. Instructional Math Coach, CT Council of Leaders of Mathematics Art in Math
Elise Archibald
 @MathSpecArch
2021年3月17日

Powerful #mathroutine to develop #computationalflexibility So excited for this!!!

CFLM_math's profile picture. Next Gen CFLM is an inquiry-based K-5 math program. Strong enough to be your core. Versatile enough to be your supplement.
Next Gen Contexts for Learning Mathematics
 @CFLM_math
2021年3月17日

We're excited #CFLM fan and Regional Math Coordinator Ryan Dent (@4ryandent) will be partnering with New Perspectives to offer a FREE workshop next Thursday--Promoting Multiplication and Division Fluency through Number Strings. Email [email protected] to register.

CFLM_math's tweet image. We&apos;re excited #CFLM fan and Regional Math Coordinator Ryan Dent (@4ryandent) will be partnering with New Perspectives to offer a FREE workshop next Thursday--Promoting Multiplication and Division Fluency through Number Strings. Email c.filo@newperspectivesonline.net to register.
1
5
10
2
0
0
0
0
0
0
未找到 "#computationalflexibility" 的结果
未找到 "#computationalflexibility" 的结果

Something went wrong.


Something went wrong.


United States Trends