Abhi Upadhyay
3 followers · 3 following · 413 views
on the atlas — 37
- [2507.16003] Learning without training: The implicit dynamics of in-context learning1 savers
- Dr. R. W. Hamming's Advice on Research14 savers
- FINNALA sleeper sofa, Gunnared beige - IKEA1 savers
- 10Gb/s Ethernet: switching to a Broadcom SFP+ module :: Giles' blog1 savers
- What the Fuck Happened to Nerds – Mr. Market2 savers
- Components of an Open Source AI Compute Tech Stack1 savers
- What we’ve learned building cloud agents · Cursor2 savers
- Encyclical Letter of His Holiness Leo XIV Magnifica Humanitas (15 May 2026)46 savers
- Why my macOS Messages badge lied to me (and the one-line fix) - Vox Silva1 savers
- How I sequenced my genome at home - home-seq2 savers
- [2411.19870] DeMo: Decoupled Momentum Optimization1 savers
- Claude Mythos Preview System Card14 savers
- Cargo Cult Science14 savers
- Running Tesla Model 3's Computer on My Desk Using Parts From Crashed Cars - bugs.xdavidhu.me2 savers
- Composer2.pdf11 savers
- Nadia Asparouhova | The independent researcher19 savers
- LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels3 savers
- TurboQuant: Redefining AI efficiency with extreme compression3 savers
- tinygrad/tinyspec: A spec for tinygrad (let's see if this works)1 savers
- flash-attention/csrc/flash_attn/src/flash_fwd_kernel.h at main · Dao-AILab/flash-attention1 savers
- PTX ISA 8.32 savers
- Questions to ask when evaluating neurotech approaches1 savers
- When AI Writes the World's Software, Who Verifies It?2 savers
- A Pragmatic Vision for Interpretability — AI Alignment Forum8 savers
- I Started Programming When I Was 7. I'm 50 Now, and the Thing I Loved Has Changed | James Randall1 savers
- Transcranial focused ultrasound for identifying the neural substrate of conscious perception - ScienceDirect1 savers
- How to Do Great Work67 savers
- Beethoven’s complete Heiligenstadt Testament | National Arts Centre1 savers
- ilaffey.com/belonging_online.html2 savers
- Teaching an LLM a Niche Diagraming Language1 savers
- LoRA Without Regret - Thinking Machines Lab37 savers
- "The World As I See It" by Albert Einstein2 savers
- Erdős Problem #124 - Discussion thread1 savers
- Circuit Tracing: Revealing Computational Graphs in Language Models20 savers
- We Induced Smells With Ultrasound11 savers
- Inside vLLM: Anatomy of a High-Throughput LLM Inference System - Aleksa Gordić11 savers
- Continuous Thought Machines2 savers
highlights — 19
This trait must be coupled with emotional commitment.
Dr. R. W. Hamming's Advice on ResearchThus an important aspect of any problem is that you have a good attack, a good starting place, some reasonable idea of how to begin.
Dr. R. W. Hamming's Advice on ResearchThus an important aspect of any problem is that you have a good attack, a good starting place, some reasonable idea of how to begin.
Dr. R. W. Hamming's Advice on Researchhe good luck to be somewhere where you are free to maintain the kind of integrity I have described, and where you do not feel forced by a need to maintain your position in the organization, or financial support, or so on, to lose your integrity. May you have that freedom
Cargo Cult ScienceMaking specifications formal and public does not resolve these questions, but it makes them explicit and auditable in a way that buried code never can.
When AI Writes the World's Software, Who Verifies It?Curiosity is the best guide. Your curiosity never lies, and it knows more than you do about what's worth paying attention to.
How to Do Great WorkCorollary: Never abandon the root node.
How to Do Great WorkYou're more likely to do great work if you're an optimist, and more likely to if you think of yourself as lucky than if you think of yourself as a victim.
How to Do Great WorkYou start with something small and evolve it, and the final version is both cleverer and more ambitious than anything you could have planned.
How to Do Great WorkSometimes you carry a question for a long time. Great work often comes from returning to a question you first noticed years before — in your childhood, even — and couldn't stop thinking about.
How to Do Great WorkSo if you're interested in a field that others think is tapped out, don't let their skepticism deter you. People are often wrong about this.
How to Do Great Workbut if you want to do great work it's an advantage to be optimistic, even though that means you'll risk looking like a fool sometimes.
How to Do Great WorkIt's not learned; it's preserved from childhood. So hold onto it.
How to Do Great WorkThat's one reason nerds have an advantage in doing great work: they expend little effort on seeming anything.
How to Do Great WorkSo just do the work and your identity will take care of itself.
How to Do Great WorkIf you're interested, you're not astray.
How to Do Great WorkBut fields aren't people; you don't owe them any loyalty. If in the course of working on one thing you discover another that's more exciting, don't be afraid to switch.
How to Do Great WorkThe three most powerful motives are curiosity, delight, and the desire to do something impressive. Sometimes they converge, and that combination is the most powerful of all.
How to Do Great WorkHowever, this does not guarantee that the local replacement model and underlying model use the same mechanisms. We can measure differences in mechanism by measuring how differently these models respond to perturbations; we refer to the extent to which perturbation behavior matches as “mechanistic faithfulness”, discussed in § 5.3 Evaluating Mechanistic Faithfulness.
Circuit Tracing: Revealing Computational Graphs in Language Models