Pinned
Bittersweet goodbye to the Farm 🌲
Successfully defended my PhD thesis 🤺 grateful for my advisors @tsachyw @sanmikoyejo and everyone I met along the way for the amazing journey at Stanford.
Berivan Isik
331 posts
Berivan Isik
@BerivanISIK
- I have joined @GoogleAI as a research scientist. I will continue to work on efficient and trustworthy AI, LLMs, safety, and privacy. Stay tuned for updates 👀
- I’ll be hosting an intern @GoogleAI in 2025 to work on the value of data for LLMs. If you’re interested, please email me your CV and a brief summary of your background. I won’t be checking DMs.
- In 2009, Google created the PhD Fellowship Program to recognize and support outstanding graduate students pursuing exceptional research in computer science and related fields. Today, we congratulate the recipients of the 2023 Google PhD Fellowship! goo.gle/3PYfLXl
- Very excited to share the paper from my last @GoogleAI internship: Scaling Laws for Downstream Task Performance of LLMs. arxiv.org/pdf/2402.04177… w/ Natalia Ponomareva, @hazimeh_h, Dimitris Paparas, Sergei Vassilvitskii, and @sanmikoyejo 1/6
- I am very excited about our new work: openreview.net/pdf?id=KeWNER6… with @RylanSchaeffer @vclecomte @sanmikoyejo @ziv_ravid @Andr3yGR @KhonaMikail @ylecun. We’ll present it in 4 @NeurIPSConf workshops: @unireps (oral), InfoCog (spotlight), @neur_reps, SSL. Details in Rylan’s tweet👇
Excited to begin announcing our #NeurIPS2023 workshop & conference papers (1/10)! 🔥🚀An Information-Theoretic Understanding of Maximum Manifold Capacity Representations🚀🔥 w/ amazing cast @vclecomte @BerivanISIK @sanmikoyejo @ziv_ravid @Andr3yGR @KhonaMikail @ylecun 1/7 - Super excited about new work Lottery Ticket Adaptation (LoTA): arxiv.org/pdf/2406.16797 We propose a sparse adaptation method that finetunes only a sparse subset of the pre-trained weights. LoTA mitigates catastrophic forgetting and enables model merging by breaking the
Excited to share Lottery Ticket Adaptation (LoTA)! We propose a sparse adaptation method that finetunes only a sparse subset of the weights. LoTA mitigates catastrophic forgetting and enables model merging by breaking the destructive interference between tasks. 🧵👇 - In our @iclr_conf 2025 paper, we study how downstream performance scales with increasing pretraining data and propose scaling laws for downstream metrics. We show that scaling behavior is significantly influenced by (1) the alignment between pretraining and downstream data and
- “Sparse Random Networks for Communication-Efficient Federated Learning” has been accepted at #ICLR2023! Code coming soon. arxiv.org/pdf/2209.15328… Looking forward to seeing many of you @iclr_conf in Rwanda.Excited to share our new work, "Sparse Random Networks for Communication-Efficient Federated Learning". arxiv.org/abs/2209.15328 1/6
- me: Vacationing across the Pacific. Totally off work. work:
- Happy to share the second paper from my @GoogleAI internship: Sandwiched Video Compression with Neural Wrappers.arxiv.org/pdf/2303.11473… The sandwich framework is more efficient than most other neural video compression methods (details below 👇). 1/3
