[ home / overboard ] [ soy / qa / raid / r ] [ craft ] [ int / pol ] [ a / an / asp / biz / mtv / r9k / tech / v / sude / x ] [ q / news / chive / rules / pass / bans / status ] [ wiki / booru / irc ]

A banner for soyjak.party

/soy/ - Soyjaks

2023 was 0.75 showers ago
Catalog
Email
Subject
Comment
File
Password (For file deletion.)

File: maxresdefault (2).jpg 📥︎ (172.95 KB, 1280x720) ImgOps

 15393809[Quote]

THE FUTURE OF EVERY NUSOIPOOPA

 15393823[Quote]

File: 846c377e326c14ad362e6f80a860a8ad57aa8fa4fc1f1….jpeg 📥︎ (84.15 KB, 1109x1346) ImgOps

first

 15393825[Quote]

File: 1770788854762d.jpg 📥︎ (73.77 KB, 600x750) ImgOps

d

 15393834[Quote]

File: 1772323282070a.gif 📥︎ (1.86 MB, 360x498) ImgOps

Nuswaglubglubs… Do I even have to say anything? The jokes are writing themselves at this point.

 15393836[Quote]

shes right :3c

 15393842[Quote]

nsense-STEM-Agent" Curriculum We curated a massive corpus of approximately 11T tokens and implemented a multi-stage training strategy. By progressively shifting the pre-training data distribution from general commonsense to complex STEM and agentic tasks, we ensure the model acquires deep cognitive abilities rather than superficial alignment. (3) Scalable Agentic Mid-training: Specifically for the agentic mid-training, we employ diverse data construction schemes to synthesize rich and varied trajectories across math, coding, and tool-use domains. This high-quality data enables the model to internalize planning and reflection behaviors effectively. Extensive evaluations show that Younsense-STEM-Agent" Curriculum We curated a massive corpus of approximately 11T tokens and implemented a multi-stage training strategy. By progressively shifting the pre-training data distribution from general commonsense to complex STEM and agentic tasks, we ensure the model acquires deep cognitive abilities rather than superficial alignment. (3) Scalable Agentic Mid-training: Specifically for the agentic mid-training, we employ diverse data construction schemes to synthesize rich and varied trajectories across math, coding, and tool-use domains. This high-quality data enables the model to internalize planning and reflection behaviors effectively. Extensive evaluations show that Younsense-STEM-Agent" Curriculum We curated a massive corpus of approximately 11T tokens and implemented a multi-stage training strategy. By progressively shifting the pre-training data distribution from general commonsense to complex STEM and agentic tasks, we ensure the model acquires deep cognitive abilities rather than superficial alignment. (3) Scalable Agentic Mid-training: Specifically for the agentic mid-training, we employ diverse data construction schemes to synthesize rich and varied trajectories across math, coding, and tool-use domains. This high-quality data enables the model to internalize planning and reflection behaviors effectively. Extensive evaluations show that Younsense-STEM-Agent" Curriculum We curated a massive corpus of approximately 11T tokens and implemented a multi-stage training strategy. By progressively shifting the pre-training data distribution from general commonsense to complex STEM and agentic tasks, we ensure the model acquires deep cognitive abilities rather than superficial alignment. (3) Scalable Agentic Mid-training: Specifically for the agentic mid-training, we employ diverse data construction schemes to synthesize rich and varied trajectories across math, coding, and tool-use domains. This high-quality data enables the model to internalize planning and reflection behaviors effectively. Extensive evaluations show that Younsense-STEM-Agent" Curriculum We curated a massive corpus of approximately 11T tokens and implemented a multi-stage training strategy. By progressively shifting the pre-training data distribution from general commonsense to complex STEM and agentic tasks, we ensure the model acquires deep cognitive abilities rather than superficial alignment. (3) Scalable Agentic Mid-training: Specifically for the agentic mid-training, we employ diverse data construction schemes to synthesize rich and varied trajectories across math, coding, and tool-use domains. This high-quality data enables the model to internalize planning and reflection behaviors effectively. Extensive evaluations show that Younsense-STEM-Agent" Curriculum We curated a massive corpus of approximately 11T tokens and implemented a multi-stage training strategy. By progressively shiftin

 15393844[Quote]

File: 1767524051558c.jpg 📥︎ (94.33 KB, 900x1000) ImgOps

D

 15393851[Quote]

File: 1769819244967g.png 📥︎ (40.48 KB, 517x504) ImgOps

>>15393834
were working on it frogpoopa

 15393883[Quote]

File: gigabump so it's not buried in bait o algo asi.mp4 📥︎ (4.48 MB, 704x1280) ImgOps

bumping this so it's not buried in bait, o algo asi

 15393897[Quote]

>>15393883
this is either a bot or a really retarded nusoi

 15393900[Quote]

Kekekekekekekekekek

 15393907[Quote]

Aryan beast

 15393946[Quote]

File: ClipboardImage (6).png 📥︎ (37.37 KB, 318x250) ImgOps


 15393948[Quote]

File: seething.gif 📥︎ (66.05 KB, 570x537) ImgOps

>>>15393883
>this is either a bot or a really retarded nusoi

 15393960[Quote]

File: 1771347330010u.png 📥︎ (216.58 KB, 588x734) ImgOps

>THE FUTURE OF EVERY NUSOIPOOPA

 15393972[Quote]

aryan gem go up

 15393976[Quote]

Bill 'gates

 15393988[Quote]

>>15393976
>>15393972
Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements. This manifesto proposes a paradigm for framing popular AI models in terms of spiking networks and polychronization, and for interpreting spiking activity as nature’s way of implementing look-up tables. This suggests a path toward converting AI models into a novel class of architectures with much smaller size yet combinatorially large encoding capacity, offering the promise of a thousandfold improvement in performance.Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements. This manifesto proposes a paradigm for framing popular AI models in terms of spiking networks and polychronization, and for interpreting spiking activity as nature’s way of implementing look-up tables. This suggests a path toward converting AI models into a novel class of architectures with much smaller size yet combinatorially large encoding capacity, offering the promise of a thousandfold improvement in performance.Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements. This manifesto proposes a paradigm for framing popular AI models in terms of spiking networks and polychronization, and for interpreting spiking activity as nature’s way of implementing look-up tables. This suggests a path toward converting AI models into a novel class of architectures with much smaller size yet combinatorially large encoding capacity, offering the promise of a thousandfold improvement in performance.Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements.

 15393998[Quote]

File: gigasurf.gif 📥︎ (6.81 MB, 560x560) ImgOps

>>>15393976
>>>15393972 (You)
>Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements. This manifesto proposes a paradigm for framing popular AI models in terms of spiking networks and polychronization, and for interpreting spiking activity as nature’s way of implementing look-up tables. This suggests a path toward converting AI models into a novel class of architectures with much smaller size yet combinatorially large encoding capacity, offering the promise of a thousandfold improvement in performance.Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements. This manifesto proposes a paradigm for framing popular AI models in terms of spiking networks and polychronization, and for interpreting spiking activity as nature’s way of implementing look-up tables. This suggests a path toward converting AI models into a novel class of architectures with much smaller size yet combinatorially large encoding capacity, offering the promise of a thousandfold improvement in performance.Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements. This manifesto proposes a paradigm for framing popular AI models in terms of spiking networks and polychronization, and for interpreting spiking activity as nature’s way of implementing look-up tables. This suggests a path toward converting AI models into a novel class of architectures with much smaller size yet combinatorially large encoding capacity, offering the promise of a thousandfold improvement in performance.Practically everything computers do is better, faster, and more power-efficient than the brain. For example, a calculator performs numerical computations more energy-efficiently than any human. Yet modern AI models are a thousand times less efficient than the brain. These models rely on larger and larger artificial neural networks (ANNs) to boost their encoding capacity, requiring GPUs to perform large-scale matrix multiplications. In contrast, the brain’s spiking neural networks (SNNs) exhibit factorially explosive encoding capacity and compute through the polychronization of spikes rather than explicit matrix–vector products, resulting in lower energy requirements.

 15394045[Quote]

File: 1703994939790.png 📥︎ (477.22 KB, 512x768) ImgOps

bumping for trans rights

 15394078[Quote]

>>15394045
bvilt for BIC

 15394081[Quote]

This post validated me so hard

 15394084[Quote]

>>15394045
The Unreasonable Effectiveness of Recurrent Neural Networks
May 21, 2015

There’s something magical about Recurrent Neural Networks (RNNs). I still remember when I trained my first recurrent network for Image Captioning. Within a few dozen minutes of training my first baby model (with rather arbitrarily-chosen hyperparameters) started to generate very nice looking descriptions of images that were on the edge of making sense. Sometimes the ratio of how simple your model is to the quality of the results you get out of it blows past your expectations, and this was one of those times. What made this result so shocking at the time was that the common wisdom was that RNNs were supposed to be difficult to train (with more experience I’ve in fact reached the opposite conclusion). Fast forward about a year: I’m training RNNs all the time and I’ve witnessed their power and robustness many times, and yet their magical outputs still find ways of amusing me. This post is about sharing some of that magic with you.

We’ll train RNNs to generate text character by character and ponder the question “how is that even possible?”

By the way, together with this post I am also releasing code on Github that allows you to train character-level language models based on multi-layer LSTMs. You give it a large chunk of text and it will learn to generate text like it one character at a time. You can also use it to reproduce my experiments below. But we’re getting ahead of ourselves.

 15394085[Quote]

Kys faggot

 15394089[Quote]

>>15394045
Indian anus tonguer

 15394111[Quote]

File: gigabump so it's not buried in bait o algo asi.mp4 📥︎ (4.48 MB, 704x1280) ImgOps

bumping this so it's not buried in bait, o algo asi

 15394112[Quote]

File: gigabump so it's not buried in bait o algo asi.mp4 📥︎ (4.48 MB, 704x1280) ImgOps

bumping this so it's not buried in bait, o algo asi



[Return][Catalog][Go to top][Post a Reply]
Delete Post [ ]
[ home / overboard ] [ soy / qa / raid / r ] [ craft ] [ int / pol ] [ a / an / asp / biz / mtv / r9k / tech / v / sude / x ] [ q / news / chive / rules / pass / bans / status ] [ wiki / booru / irc ]