41 reactions
Spotted Bumblebee at the mall today! 🤖⚡
Looks like he got lost on the way to save the world and accidentally ended up at the mall instead. 😆
#bumblebee
#transformers
Looks like he got lost on the way to save the world and accidentally ended up at the mall instead. 😆
#bumblebee
#transformers
6
5
3
1
0 reactions
Sensitivity - Positional Co-Localization in GQA Transformers
Today, I’m incredibly grateful to share a milestone that means a lot to me - my first research paper is now live on arXiv.
https://arxiv.org/abs/2604.07766
This journey wasn’t easy. It came with sleepless nights, countless iterations, debugging runs at odd hours, and pushing GPUs on runpod.io to their limits. There were moments of doubt, but also moments of deep curiosity that kept me going. Looking back, every bit of effort was worth it.
This work explores a fundamental question in GQA Transformers and led to some surprising insights around anti-localization - challenging an assumption I initially believed would hold. That’s the beauty of research: sometimes the most valuable results are the ones that prove you wrong.
This is just the beginning. Many more questions to explore, many more problems to solve.
Grateful. Motivated. Just getting started.
\#MachineLearning #Research #Transformers # AI
#technology
source
Today, I’m incredibly grateful to share a milestone that means a lot to me - my first research paper is now live on arXiv.
https://arxiv.org/abs/2604.07766
This journey wasn’t easy. It came with sleepless nights, countless iterations, debugging runs at odd hours, and pushing GPUs on runpod.io to their limits. There were moments of doubt, but also moments of deep curiosity that kept me going. Looking back, every bit of effort was worth it.
This work explores a fundamental question in GQA Transformers and led to some surprising insights around anti-localization - challenging an assumption I initially believed would hold. That’s the beauty of research: sometimes the most valuable results are the ones that prove you wrong.
This is just the beginning. Many more questions to explore, many more problems to solve.
Grateful. Motivated. Just getting started.
\#MachineLearning #Research #Transformers # AI
#technology
source
0 reactions
Sensitivity - Positional Co-Localization in GQA Transformers
Today, I’m incredibly grateful to share a milestone that means a lot to me - my first research paper is now live on arXiv.
https://arxiv.org/abs/2604.07766
This journey wasn’t easy. It came with sleepless nights, countless iterations, debugging runs at odd hours, and pushing GPUs on runpod.io to their limits. There were moments of doubt, but also moments of deep curiosity that kept me going. Looking back, every bit of effort was worth it.
This work explores a fundamental question in GQA Transformers and led to some surprising insights around anti-localization - challenging an assumption I initially believed would hold. That’s the beauty of research: sometimes the most valuable results are the ones that prove you wrong.
This is just the beginning. Many more questions to explore, many more problems to solve.
Grateful. Motivated. Just getting started.
\#MachineLearning #Research #Transformers # AI
#technology
source
Today, I’m incredibly grateful to share a milestone that means a lot to me - my first research paper is now live on arXiv.
https://arxiv.org/abs/2604.07766
This journey wasn’t easy. It came with sleepless nights, countless iterations, debugging runs at odd hours, and pushing GPUs on runpod.io to their limits. There were moments of doubt, but also moments of deep curiosity that kept me going. Looking back, every bit of effort was worth it.
This work explores a fundamental question in GQA Transformers and led to some surprising insights around anti-localization - challenging an assumption I initially believed would hold. That’s the beauty of research: sometimes the most valuable results are the ones that prove you wrong.
This is just the beginning. Many more questions to explore, many more problems to solve.
Grateful. Motivated. Just getting started.
\#MachineLearning #Research #Transformers # AI
#technology
source
