Aman Prakash (@apdotdev) / X

Aman Prakash

444 posts

Aman Prakash

@apdotdev

Research Associate @LCS2lab Prev. Research @UofSC @bitspilaniindia @reachnitt | CS'26 LLM Diffusion, MLLM Uncertainty, AI Security

Earth

linktr.ee/aman_prakash

Joined May 2023

Aman Prakash
@apdotdev
Aug 1
IIT Delhi x IBM course on ‘Responsible LLMs’ begins after their popular in-depth course on LLMs, this one covers watermarking, uncertainty, jailbreaks, interpretability, bias, etc from basics to depth i’m attending these lectures live from an IIT D. YT: youtu.be/p8T48sqIGao
Aman Prakash
@apdotdev
Jul 24
"Open source is dangerous" This is a story Frontier Labs has been selling for a while. but recently, a pre-released OpenAI model (closed-source) launched an autonomous cyberattack and an open model (GLM 5.2) came to the rescue.
From the ArtificialInteligence community on Reddit: Hugging Face says it resorted to a Chinese AI...
From reddit.com
Aman Prakash
@apdotdev
Jul 7
best feature on my smart phone for long deep work.
Aman Prakash
@apdotdev
Jul 6
x.com/apdotdev/statu…
Aman Prakash
@apdotdev
Jul 5
last month i meet 2 cofounders of @askalphaxiv (Stanford ppl & worked with Andrew NG) they were chill & very open to new ideas, i suggested few they let me test their upcoming Autoresearch (AR) feature of AlphaXiv after i explained how @karpathy AR works even for marketing
Aman Prakash
@apdotdev
Jul 5
last month i meet 2 cofounders of @askalphaxiv (Stanford ppl & worked with Andrew NG) they were chill & very open to new ideas, i suggested few they let me test their upcoming Autoresearch (AR) feature of AlphaXiv after i explained how @karpathy AR works even for marketing