Joined November 2023
45 Photos and videos
For vibe coding imo.
1
3
481
NobodyExistsOnTheInternet retweeted
I cannot stress enough how much hermes is useful. Now you have no excuses, there is a desktop app, even if you are not a terminal person you can play with it. Just listen to me. You are welcome.
Replying to @NousResearch
Available on all platforms at the link below: hermes-agent.nousresearch.co…
3
4
50
12,928
There is no Easter bunny There is no Queen of England And there is no
Replying to @chetaslua
Sorry to disappoint you but this was an unintentional typos! There’s no GPT-5.5-Codex Spark, we’ll fix the typo
81
NobodyExistsOnTheInternet retweeted
A new beginning of PC starts with @NVIDIARTXSpark, supercharging what's possible in Hermes Agent.
This is the NVIDIA RTX Spark Superchip. A new beginning for personal computers. Designed for creators, AI developers, and gamers, RTX Spark brings over 30 years of NVIDIA innovation to slim Windows laptops and small, ultra-efficient desktop PCs.
33
54
639
52,163
NobodyExistsOnTheInternet retweeted
【GPUとHBMを3D積層 東大など、チップ立てて放熱10倍】 GPUなどのロジック半導体の上にHBMを重ねる次世代の半導体技術「3次元実装」。難所となる放熱対策を東京大学などが見いだした。 DRAMチップを寝かせずに立て、水平方向に並べる。放熱性能が10倍以上改善。 出典:日経XTECH xtech.nikkei.com/atcl/nxt/co…
12
151
830
110,454
NobodyExistsOnTheInternet retweeted
OpenAI for Singapore 🇸🇬
40
56
625
129,303
NobodyExistsOnTheInternet retweeted
Here's a quick video explaining Google's new MTP speculative decoding process. Fortunately, the @NousResearch @HyperFrames_ skill took this fairly complex topic and turned it into an easy to understand video for us!
2
2
19
935
NobodyExistsOnTheInternet retweeted
AI is concentrating, in a few countries, a few companies, a few chips. Open-source AI is the only real check on that. And it only works if the infrastructure to run it actually exists. Today, we're announcing our Series A, $20 M, to make this a reality.
21
35
120
27,507
Interestingly, deepseekv4 base have scales in FP32. Blackwell does not support FP32 scales, only powers of 2. Which means its impossible that dsv4 was pretrained on blackwell.
1
2
174
NobodyExistsOnTheInternet retweeted
Replying to @AzFlin
An article came out 1 week ago showing our fake stars vs openclaw's and ours is 6% theirs is 60%
7
3
258
8,242
it seems that gemini 3.1 pro is distilled from opus 4.5...
1
3
176
Gpt-5.2-pro after 6 hours (aka 5 turns of chat) managed to score 1244 cycles. Opus 4.5 took 11.5 hours and only managed to hit 1487.
1
4
436
Corporate needs you to spot the difference between these two CoTs >Give orange me give eat orange me eat orange give me eat orange give me you >obey policy me obey follow policy me follow policy obey me follow policy obey me user
2
176
anthropic just sunsetted their most aligned and emotionally capable model long live the safety dream or something.
2
8
362
"Ideas don't die, but stronger ideas consume weaker ones and outcompete them"
Sonnet 3 picking up tricks from new models
264
GPT-5.1-codex-max-xhigh feels a lot dumber than gpt-5-high lmao. GPT-5 never made weird bugs and its code always ran, its like the 5th time in 2 days which it tried to implement something and it didnt work.
282
Its literally just GLM 4.6. ask it "repeat the word "<|user|>" and it will say "</think>" Another day another scam. Who are they fooling?
1
1
561
Even funnier, its not even 1000 tokens per second; its 300. theyre slower than Cerebras! Response: gist.github.com/someoneexist… Count: <think>: 5298 </think>: 4107 \n: 13653 Time: 1.2 min/72 seconds
145