gmagogsfm

142 posts

gmagogsfm

gmagogsfm

@gmagogsfm

ML/Finance enthusiast

Katılım Şubat 2010
307 Takip Edilen27 Takipçiler
Dirty Deeds Done Dirt Cheap
Dirty Deeds Done Dirt Cheap@kondou123452017·
@dakache01 万斯说的女性回归家庭就是曾经存在过的美好时代 今天左棍所谓的女性独立,即没有减少女性的各种问题还把生育给搞崩了,然后左棍又鼓吹人种没优劣,引入大量劣等人种吃福利 整个西方衰败
中文
2
0
0
145
李少明
李少明@dakache01·
我为什么唾弃所谓的“保守主义者”? 因为我所看到的那些保守主义者们所怀念的那个过去,从来没有在任何国家真正存在过。那是一种被重新构造出来的记忆,而不是历史。 所谓的保守主义“黄金时代”只是一个经过筛选、修饰、想象出来的过去。 真实的历史从不完美,完美的只有记忆和想象。 而保守主义者们就是活在一个想象的世界中。
中文
20
7
81
9.5K
Aidan McLaughlin
Aidan McLaughlin@aidan_mclau·
o3 is such a freak of nature
Aidan McLaughlin tweet media
English
273
179
4.4K
722.5K
difficultyang
difficultyang@difficultyang·
Trampolines are surprisingly good for 1yo babies
English
1
0
5
323
Kristof
Kristof@CoastalFuturist·
Ok, I have decided that I’m not going to cheap out on a keyboard. I use it for 40hrs a week minimum so I might as well get a good one Drop me your recs, no price limit
Kristof tweet mediaKristof tweet mediaKristof tweet mediaKristof tweet media
English
370
28
1.6K
277.1K
gmagogsfm
gmagogsfm@gmagogsfm·
@EHuanglu Why is it called 4d when it is actually 3D?
English
0
0
0
21
el.cine
el.cine@EHuanglu·
omg.. this cant be real China’s 4DV AI just dropped 4D Gaussian Splatting, you can turn 2D video into 4D with sound.. imagine.. we will be able to change camera angle, zoom in/out while watching movies 5 examples:
English
720
3.8K
35.8K
3.6M
gmagogsfm retweetledi
clem 🤗
clem 🤗@ClementDelangue·
If AI stays closed-source, proprietary and monopolistic like it is now, it will destroy lots of jobs and just make the richest companies richer and more powerful! If we open it up thanks to open science and open-source, foster competition and decentralization of value and control, it will create many more jobs and economic value than it destroys. Let’s go!
English
165
192
1.5K
154K
Community Notes & Violations
Community Notes & Violations@CNviolations·
A man built a Ferrari for his father because he couldn't afford to buy one.
English
20
22
241
34K
gmagogsfm
gmagogsfm@gmagogsfm·
@InferXai And what is utilization like with InferX?
English
1
0
0
14
InferX
InferX@InferXai·
Even the best GPU providers brag about 50% utilization. That means half the compute is still being wasted. InferX snapshot tech lets you load LLMs in 2s and pack them tightly. No warm pools. No idle.
English
1
0
1
78
ian
ian@shaoruu·
what should i add to @cursor_ai that would make u 10x more productive? feature requests, tiny nits, anything :)
English
1.4K
65
2.7K
886.1K
SimbaLee
SimbaLee@lipeng0820·
🫣毕导又在打教材的脸了 从博士的角度去细究小学初中教材里的知识点,似乎确实有些不严谨的地方。 随时AI普及的深入,未来这样的问题会越来越多还是越来越少呢?
中文
118
272
2.1K
495.8K
gmagogsfm
gmagogsfm@gmagogsfm·
@burkov Why is it 70B though? Other than experimental results, any intuitive explanation for the magic number?
English
0
0
0
34
BURKOV
BURKOV@burkov·
So far, there's no evidence that training models beyond 70B parameters provides any substantial improvement in reasoning and factual correctness. 70B parameter models can be comfortably finetuned on one 8xH100 node. Investing in domain-specific data to finetune a SOTA 70B model is today an order of magnitude, if not several orders of magnitude, more resource-efficient way of getting SOTA performance in any domain, compared to trying to train a much larger generalist model.
English
33
23
434
158.2K
Dia Kharrat
Dia Kharrat@dkharrat·
@rakyll It’s time for Google to move from C++ to Rust.
English
4
0
32
7K
gmagogsfm
gmagogsfm@gmagogsfm·
@NielsRogge The decision of which expert to activate tends to be per token, so to process one prompt you'd need fast access to many experts, not just one . If weights are swapped in and out, the latency will be unbearable
English
0
0
0
78
Niels Rogge
Niels Rogge@NielsRogge·
I'm a noob when it comes to running MoEs on GPUs. Can you store the model's expert parameters on disk and just load the activated ones on GPU?
English
24
2
171
16.1K
gmagogsfm
gmagogsfm@gmagogsfm·
@BBITJungle @onenewbite Market always recovers until it doesn't any more. 我不确定这次是不是会不一样,但是盲目以有限的过去经验推导规律并不是正确的逻辑,反而是类似信仰的伪科学
中文
2
0
0
414
BabyInTheJungle
BabyInTheJungle@BBITJungle·
一点都不慌是假的 但是每次慌的时候就加一点仓 历史证明stay in the market is always the best strategy; 总有人会说: this time is different 但是历史证明 market always recovers 那些一波操作猛如虎的人 最终都会被证明是二百五 🤣 有的是时间等 短的几个月 长的话几年 但是不要用杠杆 不要用你短期要用的钱投资
中文
2
1
26
7.1K
一口新饭
一口新饭@onenewbite·
真的不理解大家在慌什么。。投资者请讲讲你担心的理由和操作。对我来说,我攒的现金终于有些机会可以用了。
中文
69
5
305
257.7K
gmagogsfm
gmagogsfm@gmagogsfm·
@OpenAI @sama Is ChatGPT experiencing problems? I am hitting this error for request
gmagogsfm tweet media
English
0
0
0
18
Tom Warren
Tom Warren@tomwarren·
AMD’s Radeon RX 9070 series launch looks like it’s going to be anything but a paper one. Overclockers UK says it has “several thousand cards in stock spanning across Sapphire, PowerColor and Asrock all of which have 1000+ in stock of each brand” forums.overclockers.co.uk/threads/resist…
Tom Warren tweet mediaTom Warren tweet media
English
28
50
851
61.4K
𝗦𝘁𝗲𝘃𝗲 𝕏
𝗦𝘁𝗲𝘃𝗲 𝕏@st7evechou·
馋死了。🤩🤩🤩 不得不说啊,现在国产的键盘又卷价格又卷配置,做工也不错!
𝗦𝘁𝗲𝘃𝗲 𝕏 tweet media𝗦𝘁𝗲𝘃𝗲 𝕏 tweet media
中文
37
7
260
92.6K
gmagogsfm
gmagogsfm@gmagogsfm·
@LisaSu Please don't miss this opportunity!
English
0
0
0
39