Hacker News: sabareesh

New comment by sabareesh in "Claude Opus 4.7"

sabareesh — Thu, 16 Apr 2026 18:42:12 +0000

Based on last few attemts on claude code to address a docker build issue this feels like a downgrade

New comment by sabareesh in "OpenAI's latest repo has Claude as the third top contributor"

sabareesh — Wed, 25 Mar 2026 15:11:17 +0000

Codex usually dont add itself as contributor so this is misleading .

New comment by sabareesh in "Nvidia greenboost: transparently extend GPU VRAM using system RAM/NVMe"

sabareesh — Wed, 18 Mar 2026 23:30:08 +0000

I wish it provided benchmark comparing Direct RAM offload vs CPU offload vs Full VRAM

Teaching Qwen3-4B to Trade: From Hold-Collapse to +9.4% Returns

sabareesh — Fri, 13 Mar 2026 22:19:53 +0000

Article URL: https://sabareesh.com/posts/trading-llm/

Comments URL: https://news.ycombinator.com/item?id=47370723

Points: 2

# Comments: 0

DeepWiki and Increasing Malleability of Software

sabareesh — Sun, 15 Feb 2026 04:54:36 +0000

Article URL: https://twitter.com/karpathy/status/2021633574089416993

Comments URL: https://news.ycombinator.com/item?id=47021176

Points: 1

# Comments: 0

New comment by sabareesh in "Self Driving Car Insurance"

sabareesh — Fri, 30 Jan 2026 17:27:05 +0000

Tesla have their own Insurance product which is already very competitive compared to other providers. Not sure if lemonade can beat them . Tesla's insurance product has similar objective in place already where it rewards self driving over manual driving.

New comment by sabareesh in "Stop Doom Scrolling, Start Doom Coding: Build via the terminal from your phone"

sabareesh — Tue, 06 Jan 2026 20:26:09 +0000

I am looking for some open source terminal for iphone .I have code server running which i can just use terminal from vs code on safari

New comment by sabareesh in "I switched from VSCode to Zed"

sabareesh — Mon, 05 Jan 2026 17:27:26 +0000

Sorry to disappoint. But purely codex and claude code

New comment by sabareesh in "I switched from VSCode to Zed"

sabareesh — Mon, 05 Jan 2026 14:54:26 +0000

I have switched to terminal

New comment by sabareesh in "IQuest-Coder: A new open-source code model beats Claude Sonnet 4.5 and GPT 5.1 [pdf]"

sabareesh — Sat, 03 Jan 2026 05:55:56 +0000

TL;DR is that they didn't clean the repo (.git/ folder), model just reward hacked its way to look up future commits with fixes. Credit goes to everyone in this thread for solving this: https://xcancel.com/xeophon/status/2006969664346501589

(given that IQuestLab published their SWE-Bench Verified trajectory data, I want to be charitable and assume genuine oversight rather than "benchmaxxing", probably an easy to miss thing if you are new to benchmarking)

https://www.reddit.com/r/LocalLLaMA/comments/1q1ura1/iquestl...

New comment by sabareesh in "Show HN: Stop Claude Code from forgetting everything"

sabareesh — Tue, 30 Dec 2025 00:10:47 +0000

Non starter for us, we cant ship propriety data to a third party servers.

New comment by sabareesh in "Gemini 3 Flash: Frontier intelligence built for speed"

sabareesh — Thu, 18 Dec 2025 00:07:35 +0000

this has one of the worse score in AA-Omniscience Hallucination Rate

New comment by sabareesh in "Gemini 3 Flash: Frontier intelligence built for speed"

sabareesh — Thu, 18 Dec 2025 00:07:09 +0000

Nope lower is better compared to recent open ai models this is bad. I am looking at AA-Omniscience Hallucination Rate

New comment by sabareesh in "Gemini 3 Flash: Frontier intelligence built for speed"

sabareesh — Wed, 17 Dec 2025 23:49:39 +0000

Watch out these model are hallucinating lot more https://artificialanalysis.ai/evaluations/omniscience?omnisc...

New comment by sabareesh in "The Big Vitamin D Mistake [pdf] (2017)"

sabareesh — Tue, 09 Dec 2025 22:14:14 +0000

So is 10,000 IU of daily does ok ?

One point I made that didn't come across: Ilya

sabareesh — Fri, 28 Nov 2025 20:38:28 +0000

Article URL: https://twitter.com/ilyasut/status/1994424504370581726

Comments URL: https://news.ycombinator.com/item?id=46082556

Points: 7

# Comments: 0

Show HN: MCP Compact – Middleware to trim noisy browser/DOM/tool outputs

sabareesh — Thu, 20 Nov 2025 02:50:14 +0000

Modern MCP agents often blow past context limits because tools return huge blobs (DOM dumps, network traces, screenshots, etc.). MCP Compact is a small proxy that sits between your agent and the upstream MCP server, forwards every tool call, and uses an LLM to summarize the response according to per-tool rules (token budget, fields to preserve/drop). It supports StreamableHTTP, tracks token usage, and can reconnect sessions, so you can keep agents focused without rewriting agent code or tools.

Comments URL: https://news.ycombinator.com/item?id=45988299

Points: 1

# Comments: 0

New comment by sabareesh in "When Tesla's FSD works well, it gets credit. When it doesn't, you get blamed"

sabareesh — Mon, 10 Nov 2025 00:03:54 +0000

Technically you kind of get this in Nevada when using Tesla insurance and if you drive 100 % FSD. If you drive manually you are pretty much doxed for random Front collision Warning which is super sensitive

New comment by sabareesh in "Karpathy on DeepSeek-OCR paper: Are pixels better inputs to LLMs than text?"

sabareesh — Wed, 22 Oct 2025 22:18:52 +0000

It might be that our current tokenization is inefficient compared to how well image pipeline does. Language already does lot of compression but there might be even better way to represent it in latent space

New comment by sabareesh in "Claude Code on the web"

sabareesh — Mon, 20 Oct 2025 21:33:34 +0000

Similar feeling. Seems it is good at certain things and if something doesnt work it want to do things simply and in turn becomes something that you didnt ask for and certain times opposite of what you wanted. On the other hand with codex certain time you feel the AGI but that is like 2 out of 10 sessions. This is primarily may be due to how complete the prompt and how well you define the problems.