Hacker News: viksit

New comment by viksit in "I have officially retired from Emacs"

viksit — Tue, 28 Apr 2026 18:10:52 +0000

> heavily-customized and terminal-heavy setup

this exactly. most people can’t set it up that well.

New comment by viksit in "Personal Computer by Perplexity"

viksit — Wed, 11 Mar 2026 22:31:21 +0000

underrated comment haha. made my day

New comment by viksit in "A lot of the Moltbook stuff is fake"

viksit — Sat, 31 Jan 2026 18:07:45 +0000

I think the point here is the bot didn’t come up with the idea on its own, but was seeded by a human with an entrenched interest / experience in that space. The PR around Moltbook has been that Agents are coming up with these ideas through some sort of swarm interaction with each other, and developing novel, emergent capabilities — which is not true.

New comment by viksit in "Tongyi DeepResearch – open-source 30B MoE Model that rivals OpenAI DeepResearch"

viksit — Sun, 02 Nov 2025 18:53:33 +0000

this is a great question. what are the main use cases that you have for this? i’ve been working on a library for something similar and exposing it via an mcp interface. would love to pick your brain on this (@viksit on twitter)

New comment by viksit in "How hard do you have to hit a chicken to cook it? (2020)"

viksit — Sat, 11 Oct 2025 17:34:38 +0000

Yes! thanks for the memory haha.

New comment by viksit in "Altered states of consciousness induced by breathwork accompanied by music"

viksit — Thu, 28 Aug 2025 13:35:43 +0000

you definitely succeeded in your humorous endeavors ;) i snorted haha

New comment by viksit in "Replacing Agent Prompt Templates with Trainable Control Modules in DSPy"

viksit — Thu, 07 Aug 2025 15:26:01 +0000

In my fourth post in the series, I tackle how to make multi-step agent workflows learn behavior from data. Most agents today rely on vibes: prompt tuning, hand-written templates, and hope(!). This post is about replacing that with metrics and optimization.

Each branch in the workflow learns how to behave, not just where to route. I show how to set up a reward, plug in an optimizer, and treat agent behavior as something you can tune like a model.

Replacing Agent Prompt Templates with Trainable Control Modules in DSPy

viksit — Thu, 07 Aug 2025 15:26:01 +0000

Article URL: https://viksit.substack.com/p/behavioral-optimization-for-multi

Comments URL: https://news.ycombinator.com/item?id=44825724

Points: 2

# Comments: 1

New comment by viksit in "GEPA: Reflective prompt evolution can outperform reinforcement learning"

viksit — Fri, 01 Aug 2025 03:00:42 +0000

mostly aligned on this. couple of thoughts:

- raw accuracy is now a "vanity" metric. so the benchmarks need to get more sophisticated, and i think they're going to have to be far more task specific than hotpot or hover. they've become like the mnist of multi hop.

- in my use of MIPROv2 and SIMBA, I see a fair amount of improvements for multi hop tasks (published some of these on hn before). I'm going to try GEPA and see how it performs. so I think we're at the start of what I would call "meta learning".. tuning across a huge search surface rather than tweaking one prompt. hyper param search for higher dim spaces.

- tokens burned should be a reported result

New comment by viksit in "GEPA: Reflective prompt evolution can outperform reinforcement learning"

viksit — Fri, 01 Aug 2025 02:44:28 +0000

vJEPA models, lecun's approach towards world models that have been derided by a lot of naysayers. (personally I think thats the way to go)

New comment by viksit in "GEPA: Reflective prompt evolution can outperform reinforcement learning"

viksit — Thu, 31 Jul 2025 13:11:07 +0000

they’ve already written one! see omar’s x account for details!

New comment by viksit in "Optimizing Tool Selection in LLM Workflows(Part 2): A DSPy and PyTorch Benchmark"

viksit — Wed, 23 Jul 2025 20:58:26 +0000

Based on the feedback from my last HN post on differentiable routing, I ran a follow-up benchmark: local RNN vs GPT-4o for tool selection in LLM workflows.

Same accuracy, 40% lower cost. Appreciate all the suggestions, this post builds on them.

Optimizing Tool Selection in LLM Workflows(Part 2): A DSPy and PyTorch Benchmark

viksit — Wed, 23 Jul 2025 20:58:26 +0000

Article URL: https://viksit.substack.com/p/evaluation-tool-selection-in-llms

Comments URL: https://news.ycombinator.com/item?id=44663980

Points: 1

# Comments: 1

New comment by viksit in "Differentiable Programming for Learnable Graphs: Optimizing LLM Workflows W DSPy"

viksit — Sun, 13 Jul 2025 23:14:55 +0000

Following up on my last post about optimizing tool selection with differentiable programming, I’ve been thinking about how to extend those ideas to full agent workflows. This post shares some early experiments using DSPy to optimize routing and structure end-to-end for a sample customer service agent workflow. Feedback welcome!

Differentiable Programming for Learnable Graphs: Optimizing LLM Workflows W DSPy

viksit — Sun, 13 Jul 2025 23:14:55 +0000

Article URL: https://viksit.substack.com/p/differentiable-programming-for-learnable

Comments URL: https://news.ycombinator.com/item?id=44554665

Points: 2

# Comments: 1

New comment by viksit in "Optimizing Tool Selection for LLM Workflows with Differentiable Programming"

viksit — Tue, 08 Jul 2025 16:02:39 +0000

would you have a link?

New comment by viksit in "Optimizing Tool Selection for LLM Workflows with Differentiable Programming"

viksit — Tue, 08 Jul 2025 16:02:30 +0000

there's a world where the model could infer that as well!

New comment by viksit in "Optimizing Tool Selection for LLM Workflows with Differentiable Programming"

viksit — Tue, 08 Jul 2025 16:01:46 +0000

yes, AFAIK right now, there are no easy ways of "slimming" context because no one knows what it should be or how.

New comment by viksit in "Optimizing Tool Selection for LLM Workflows with Differentiable Programming"

viksit — Tue, 08 Jul 2025 16:01:13 +0000

great q. thats coming up as a future post in the series.

New comment by viksit in "Optimizing Tool Selection for LLM Workflows with Differentiable Programming"

viksit — Tue, 08 Jul 2025 16:00:56 +0000

this is really interesting! yes, its my substack.

also, if you're down, love to connect and talk more about what use cases / techniques you're using. I'm @viksit on X dms if that works.