Nasim Rahaman
@nasimrahaman.bsky.social
👨🍳 @ Tiptree Systems. Previously, deep learning & espresso slurping @ Mila and Max-Planck Institute for Intelligent Systems. Before that, physics and more deep learning @ Uni Heidelberg.
📍Berlin
📍Berlin
The purpose of a document is to help an LLM answer questions about said document.
July 6, 2025 at 3:33 PM
The purpose of a document is to help an LLM answer questions about said document.
Re: Dario’s post (darioamodei.com/on-deepseek-...) — the current state of the US has left me scared shitless about a unipolar world. A single, powerful pole with compromised institutions is the worst possible outcome for human civilization.
Dario Amodei — On DeepSeek and Export Controls
On DeepSeek and Export Controls
darioamodei.com
January 29, 2025 at 10:36 PM
Re: Dario’s post (darioamodei.com/on-deepseek-...) — the current state of the US has left me scared shitless about a unipolar world. A single, powerful pole with compromised institutions is the worst possible outcome for human civilization.
I’m low key disappointed r1 doesn’t swear in its CoT.
Tired: “Wait…”
Wired: “hol the fuck up”
Tired: “Wait…”
Wired: “hol the fuck up”
January 25, 2025 at 11:10 PM
I’m low key disappointed r1 doesn’t swear in its CoT.
Tired: “Wait…”
Wired: “hol the fuck up”
Tired: “Wait…”
Wired: “hol the fuck up”
An underappreciated takeaway from Gwern’s recent post (linked) is that LLMs are like vessels of human culture. They’re distribution channels for values.
gwern.net/creative-ben...
gwern.net/creative-ben...
Towards Benchmarking LLM Diversity & Creativity
Discussion of possible tasks to measure LLM capabilities in soft ‘creative’ tasks like brainstorming or editing, to quantify failures in creative writing domains.
gwern.net
January 17, 2025 at 10:56 PM
An underappreciated takeaway from Gwern’s recent post (linked) is that LLMs are like vessels of human culture. They’re distribution channels for values.
gwern.net/creative-ben...
gwern.net/creative-ben...
Finished the first season of Severance. Brb implementing memory for my LLM agents.
January 8, 2025 at 10:50 PM
Finished the first season of Severance. Brb implementing memory for my LLM agents.
Claude is something special.
December 16, 2024 at 10:39 PM
Claude is something special.
Very cool work from Meta AI: Large Concept Models. Idea: autoregress in the space of sentence level representations.
I think an interesting next step would be to layer this on conventional LLMs / token prediction models.
Here’s how that could work: ⤵️
I think an interesting next step would be to layer this on conventional LLMs / token prediction models.
Here’s how that could work: ⤵️
December 14, 2024 at 5:43 AM
Very cool work from Meta AI: Large Concept Models. Idea: autoregress in the space of sentence level representations.
I think an interesting next step would be to layer this on conventional LLMs / token prediction models.
Here’s how that could work: ⤵️
I think an interesting next step would be to layer this on conventional LLMs / token prediction models.
Here’s how that could work: ⤵️
Reposted by Nasim Rahaman
Microsoft just released a tool that lets you convert Office files to Markdown. Never thought I'd see the day.
Google also added Markdown export to Google Docs a few months ago.
github.com/microsoft/markitdown
Google also added Markdown export to Google Docs a few months ago.
github.com/microsoft/markitdown
GitHub - microsoft/markitdown: Python tool for converting files and office documents to Markdown.
Python tool for converting files and office documents to Markdown. - microsoft/markitdown
github.com
December 13, 2024 at 8:25 PM
Microsoft just released a tool that lets you convert Office files to Markdown. Never thought I'd see the day.
Google also added Markdown export to Google Docs a few months ago.
github.com/microsoft/markitdown
Google also added Markdown export to Google Docs a few months ago.
github.com/microsoft/markitdown
My MSc advisee (& gang) cooked.
tl:dr — a cute technique to get machines to be more creative.
🧵⤵️
tl:dr — a cute technique to get machines to be more creative.
🧵⤵️
December 13, 2024 at 9:41 AM
My MSc advisee (& gang) cooked.
tl:dr — a cute technique to get machines to be more creative.
🧵⤵️
tl:dr — a cute technique to get machines to be more creative.
🧵⤵️
If you aren’t already doing this: share that pdf with Claude.
December 7, 2024 at 10:50 PM
If you aren’t already doing this: share that pdf with Claude.
Plug for maybe the smoothest take on agents that I’ve seen: arxiv.org/abs/2007.05300 by @rovelli.bsky.social.
🧵 on my read:
🧵 on my read:
Agency in Physics
I discuss three aspects of the notion of agency from the standpoint of physics: (i) what makes a physical system an agent; (ii) the reason for agency's time orientation; (iii) the source of the inform...
arxiv.org
November 23, 2024 at 4:57 AM
Plug for maybe the smoothest take on agents that I’ve seen: arxiv.org/abs/2007.05300 by @rovelli.bsky.social.
🧵 on my read:
🧵 on my read: