To everyone: The busy state of DeepSeek is not a server issue; it is because DeepSeek is under continuous large-scale cyber attacks! And all the attacking IPs originate from the United States.
Well they getting a lot of hype these few weeks make sense they grtting overload, the choicenwe could have are waiting for them to get more huawei chip for inference. They couldn't get nvidia as easy as previously i belive so yeah
The reason deepseek being attacked is not about the models, nor the data used by training. But the training method it used, which lower the hw requirement by 80% or more.
Question ⁉️ when you train the model how do you choose how many parameters are activated each pass ! And how do they know . Is this the training settings ? Should we be fine tuning a higher amount of parameters to make a better response ?
If your just using chat and not API then id suggest kimi ai if deepseek is still under attack. Its free, also a reasoning model but its multimodal and has search.
@@MS-wz9jmI use it via API with Cline through OpenRouter, but besides I use it with chat for side small tasks simultaniously, because even as cheap as it is, it just bothers me paying for the same product I can get completely for free... Although, really, if we look at performance, they're not the same product, so...
To everyone: The busy state of DeepSeek is not a server issue; it is because DeepSeek is under continuous large-scale cyber attacks! And all the attacking IPs originate from the United States.
"Source: Your mom."
Hug of death. 🤗💀
@@madbaron228 "Source: common sense"
maybe cloudflare could help
pshhh, every other company can deal with bots. Deepseek has weak Infrastructure engineering
Just fix the busy server first ah.. I'm one of the earliest users of deepseek and never had this problem, now I'm stress
Same, evwrything went to crap when the full R1 came out. I miss the R1 lite days...
Run it locally
@@cbgaming08then it's not worth, useless you have 24gb vram+ GPU or 128 gb ram
Well they getting a lot of hype these few weeks make sense they grtting overload, the choicenwe could have are waiting for them to get more huawei chip for inference.
They couldn't get nvidia as easy as previously i belive so yeah
same situation bro 🙂
Deepseek is killing it on all fronts, there's no other comment to make.
CRAZY! fast update bro.
The reason deepseek being attacked is not about the models, nor the data used by training. But the training method it used, which lower the hw requirement by 80% or more.
Question ⁉️ when you train the model how do you choose how many parameters are activated each pass ! And how do they know . Is this the training settings ? Should we be fine tuning a higher amount of parameters to make a better response ?
Where do I get more of the intro song?
just play it as loop
Could you actually this for non-vision tasks? I want to see how it compares to the R1 distillations.
Great content. ❤
Hope they improve their servers.
All i get in response is that the server is busy, all the time
💯
It's under constant cyber attack from the US and it's allies.
On a lucky day I get up to 2 requests within a session 😂
If your just using chat and not API then id suggest kimi ai if deepseek is still under attack. Its free, also a reasoning model but its multimodal and has search.
@@MS-wz9jmI use it via API with Cline through OpenRouter, but besides I use it with chat for side small tasks simultaniously, because even as cheap as it is, it just bothers me paying for the same product I can get completely for free... Although, really, if we look at performance, they're not the same product, so...
Love Deepseek. Now everyone stop using so I can actually use it. :(
It would be nice if they added it to ollama
There's way to download and make it a usable model on ollama
ClosedAI and MS don't like any true rivals
Is there any way to upload your on PDF file to the deep Seek and ask it question about the PDF??
Is it CRAZY??? Or is it actually INNSAAAANNNEEE!!!!!!!11
¡Qué chulo!.
En mi opinión a R1 solo le falta Visión y la capacidad de mover y comunicarse con robot vía wifi o 5G.
link 2 deepseek space in HF?
Deepseek are too busy to accept my data. I using deepseek since it come out around 2 yrs ago.
Wow fast! 😊
AWESOME!
Deepseek is 99% unusable currently.
I hope they fix the API problems soon
Deepseek need a new server💔
why does bro talk like a snail :( doesnt even get better at 2x speed
hi how can i contact you?
Chatgpt and others crying in corner and tells deepsake stole from them 😂
Talk about new github copilot Agent in the VScode Inside
I used it today for some complex task, I really enjoyed the fact that there were no errors on the code it generated, though it's a bit slow..
yeahhhhh
can i have i break I am tired
Hahaha 😂
Focus on these days. This will feel like a lazy river within ~6 months
sure, go on, have a break.