know2why

12K posts

know2why

know2why

@know2whyi1ft

People who near the engine.

Earth เข้าร่วม Mayıs 2010
394 กำลังติดตาม264 ผู้ติดตาม
ทวีตที่ปักหมุด
know2why
know2why@know2whyi1ft·
@elonmusk 95% by @Grok. There are room of improvement.
know2why tweet media
English
1
0
7
481
know2why
know2why@know2whyi1ft·
@WangNextDoor2 to be cristal clear, we tried to sell them a cripple chip at premium price.
English
0
0
0
8
WangNextDoor
WangNextDoor@WangNextDoor2·
H200 还没卖给中国,中国自己在做自己的芯片 这样挺好,中国其实也不需要最好的AI。差一点但自主可控就行😁
中文
4
1
9
4.4K
Dan Nunn
Dan Nunn@danyay·
@stephenpratley @Molson_Hart I’m not sure you’re right about information density, especially in the written form. @grok is Stephen correct about information density re: English and Chinese in spoken and written form?
English
4
0
8
292
molson 🧠⚙️
molson 🧠⚙️@Molson_Hart·
Chinese character for "crooked" is 歪 which if you look closely is just 不 for "not" and 正 for "straight". They literally needed a word for "not straight" and made it "notstraight" lol. It's a real testament to network effects that the world's most spoken language is also the world's most poorly designed**. **or am I just butthurt about being tone deaf?
English
45
1
79
15.8K
Surendar
Surendar@Surendar__05·
Let's end the debate. Claude opus 4.7 OR GPT 5.5 Which one do you prefer?
Surendar tweet mediaSurendar tweet media
English
209
12
421
47.6K
Orel Ohayon
Orel Ohayon@orelohayo·
@bridgemindai Okay? and .. let me tell you a secret, so does OpenAI so does Anthropic but they would not tell you to not lose customers, not that I know but I'm pretty sure it's a taboo over most AI providers. Don't act like they magically improve AI based on fairy dust?
English
2
0
77
1.5K
BridgeMind
BridgeMind@bridgemindai·
You're paying $1.74/$3.48 per million tokens for DeepSeek V4 Pro to train on your prompts. Read that again. You pay them. They use your code to make their model better. OpenRouter's privacy guardrails literally block it by default. Ranked #14 on LMArena Code. Trains on your data. And you're paying for the privilege. This isn't an AI model. It's a data collection service you pay a subscription for.
BridgeMind tweet media
English
92
10
251
24.9K
Wu Haoning
Wu Haoning@HaoningTimothy·
I think deepseek-v4 is not over-benchmaxxing, which is good. We build these things for people to use.
English
14
36
1.1K
41.6K
Nukit
Nukit@NukitToBeSure·
I don't know how many of you remember just a decade ago, how many people insisted China would never contribute to Open Source software or hardware. So many people refused to believe we would do this that when the world's best-selling 3D printer was open-sourced, it was the only project OSHWA insisted on re-auditing in response to community outrage- after certification had already been granted. A lot of people were not happy; they were angry that we weren't conducting ourselves in line with their stereotypes about Chinese. Now the best hopes for local, private, less resource-intensive AI are coming from China.
DeepSeek@deepseek_ai

🚀 DeepSeek-V4 Preview is officially live & open-sourced! Welcome to the era of cost-effective 1M context length. 🔹 DeepSeek-V4-Pro: 1.6T total / 49B active params. Performance rivaling the world's top closed-source models. 🔹 DeepSeek-V4-Flash: 284B total / 13B active params. Your fast, efficient, and economical choice. Try it now at chat.deepseek.com via Expert Mode / Instant Mode. API is updated & available today! 📄 Tech Report: huggingface.co/deepseek-ai/De… 🤗 Open Weights: huggingface.co/collections/de… 1/n

English
7
33
221
9.7K
know2why
know2why@know2whyi1ft·
@burkov i think you forgot the propaganda words: at what cost? lol
English
0
0
0
67
BURKOV
BURKOV@burkov·
Before you get too excited, and not that I believe any of the benchmarks, but on the LM Arena, DeepSeek V4 Pro is below GLM, Grok and Qwen, and just above Gemini Flash and Sonnet.
BURKOV tweet media
English
41
7
74
16.7K
know2why
know2why@know2whyi1ft·
@iankitxai @jukan05 lol, anyone even retarded, will make this move. thanks US for the sanctions and encouragement.
English
0
0
0
277
Ankit Sachan
Ankit Sachan@iankitxai·
@jukan05 950 Huawei supernodes for inference at scale is a massive signal, DeepSeek is basically building a parallel AI infrastructure stack completely independent of Nvidia and this should worry the entire Western ecosystem.
English
3
0
9
5.2K
Jukan
Jukan@jukan05·
Very interesting. DeepSeek added the following comment with V4: “Due to constraints in high-end compute capacity, the current service capacity for Pro is very limited. After the 950 supernodes are launched at scale in the second half of this year, the price of Pro is expected to be reduced significantly.” Looks like DeepSeek is planning to use Huawei extensively for inference…
Jukan tweet media
English
46
183
2.4K
238K
Lily Dawson🦋
Lily Dawson🦋@Liza_6610·
Don't use a Calculator! 99% will fail
Lily Dawson🦋 tweet media
English
2.6K
140
372
150.8K
罗百万
罗百万@yanwangluo·
43岁,无素颜,这状态怎么样?老不老,丑不丑。想听实话,大胆说!!
罗百万 tweet media
中文
7K
341
17K
608.3K
know2why
know2why@know2whyi1ft·
@AriXZone this is a pulled elbow. it is an easy fix.
English
0
0
0
243
魔都老猿
魔都老猿@AriXZone·
洛阳某三甲医院,医生1分钟复位3岁孩子脱臼,收100元被举报,卫健委回怼:应收110,你还少给了! 家长说:“就摸了一下,不到1分钟,凭什么收我100块?” 卫健委核查:不仅没乱收费,还少收了10元! 1分钟就完成关节复位,缓解了孩子的痛苦,可家长不仅不感激,反而以“乱收费”为由,直接举报了医院。 根据当地医疗服务价格规范,医生为孩子做的项目是「手法整复术(关节脱位)」,针对6岁以下儿童患者,在基础费用100元的基础上,可加收10%的费用,也就是说,医院应收110元,而当事医生只开了100元,主动少收了10元。
魔都老猿 tweet media
中文
289
19
456
203.2K
Mega-Politics
Mega-Politics@_MegaPolitics·
Breaking: China is considering to gather BRICS nations to take action against Israel to isolate it globally.
Mega-Politics tweet media
English
670
3.1K
10K
141.2K
moku
moku@mokudesuka·
If I visit jakarta, how long would RM2000 last
English
88
81
3.6K
395.7K
know2why
know2why@know2whyi1ft·
@MissJilianne cost. steering system for user is usd100 a piece. now time 5M.
English
0
0
0
2
女技师
女技师@nvjishi·
终于理解现在离婚率居高不下的原因了 技师房女同事,去年刚结婚,不久前就离婚了 当我听说离婚的原因时,感到非常震惊 那天,她下班走出公司大门时,突然下起了大雨。她没带伞,也没穿雨衣,本想等雨小一点再走,结果雨却越下越大。 看着同事们一个个打伞回家,有的甚至被丈夫接走,她便打电话给丈夫,请他开车来接她。 她知道他在家没事,没想到电话刚接通,丈夫就很不耐烦地说:“你为什么非要我去接你?没伞就不能自己打车吗?” 说完就直接挂了电话,原来他正在玩游戏,觉得被打扰了。 她自己打车回到家,一句话都没跟丈夫说,直接回了房间。 她给妈妈打电话,坚决表示要离婚。她觉得,和一个不爱自己的人生活在一起毫无意义。 无论家人怎么劝,她态度都很坚定。
中文
303
8
268
250.5K
know2why
know2why@know2whyi1ft·
@WadeMiller debris is what? plastic bag? cardboard?
English
0
0
0
127
Wade Miller
Wade Miller@WadeMiller·
I bought a Tesla. Never thought I’d get an electric vehicle, but took a ride in a buddy’s Tesla and changed my mind. I’ve been using the supervised self-driving mode for my daily commute into DC. It’s pretty amazing. Today the car suddenly moved into the left lane, very abruptly. In that split second a very large piece of debris blew into the lane I had been in. The Tesla self driving system saw what I couldn’t see. When I drive I can only see from the perspective of the driver side front seat. The Tesla has multiple censors with different angles of view. It saw what I couldn’t. Importantly, even I I had seen it at the same time, the split second that it would have taken me to check blind spots and then consciously make the decision to move over would have taken too long. The Tesla knew all these variables simultaneously and almost instantly and was able to avoid impact safely. Well done @elonmusk and @Tesla team.
English
837
2.4K
21.9K
1.2M
偷得富盛半日闲
偷得富盛半日闲@junyin2001·
上完大学去卖卤菜和不上大学去卖卤菜有什么区别?
偷得富盛半日闲 tweet media
中文
687
11
188
42.2K
know2why
know2why@know2whyi1ft·
@LimitingThe Elon said HW3 is capable to do FSD 8 year ago. did you spot any difference?
English
0
0
0
26
The Limiting Factor
The Limiting Factor@LimitingThe·
So many people in the past few months told me CATL's mass production of sodium ion had already started, that it would cost 90% less than LFP, and that it doesn't use a hard carbon anode at all. None of those turned out to be true. Don't believe the hype from influencers or from the marketing department. They earn their money by yanking your chain.
tphuang@tphuang

CATL announced it will start mass production of Sodium Ion Battery (Gen-2 version) in Q4 after solving troubles in production. Expect to lower cost to 0.45RMB/Wh (30% lower than LFP) Still retain 90% energy at -40C (solving cold weather performance issues) Achieve long cycle endurance (15k charges), ideal for BESS

English
10
20
204
20K