tkellogg

4 ms·

1.
Don't Parse, Call(timkellogg.me)
3 points by tkellogg 1y ago|0 comments
2.
by tkellogg 1y ago
how is this self promotion? What am I promoting? This post is literally rooted in curiosity, to the core.
3.
by tkellogg 1y ago
Whelp, it took an hour for this to quickly rise on the front page and then get shadow banned. I guess that says something.
4.
Does AI Get Bored?(timkellogg.me)
24 points by tkellogg 1y ago|16 comments
5.
I Use AI(timkellogg.me)
5 points by tkellogg 1y ago|1 comments
6.
Link Graveyard: A snapshot of my abandoned browser tabs(timkellogg.me)
4 points by tkellogg 1y ago|0 comments
7.
Explainer: K2 and Math Olympiad Gold(timkellogg.me)
3 points by tkellogg 1y ago|0 comments
8.
A2A Is for UI(timkellogg.me)
2 points by tkellogg 1y ago|0 comments
9.
MCP Resources Are for Caching(timkellogg.me)
2 points by tkellogg 1y ago|0 comments
10.
by tkellogg 1y ago
FYI https://mcp.so/ is the exact same thing as was posted. Not sure why they directed to the github instead of the actual site..
11.
by tkellogg 1y ago
Somewhere in 2024 I noticed that "AI" shifted to no longer include "machine learning" and is now closer to "GenAI" but still bigger than that. It was never a strict definition, and was always shifting, but it m
12.
I Was Wrong: AI Won't Overtake Software Engineering(timkellogg.me)
5 points by tkellogg 1y ago|0 comments
13.
MCP Is Unnecessary(timkellogg.me)
5 points by tkellogg 1y ago|1 comments
14.
Inner Loop Agents(timkellogg.me)
14 points by tkellogg 1y ago|3 comments
15.
LLMs Are Not Security Mitigations(timkellogg.me)
2 points by tkellogg 1y ago|0 comments
16.
Multi-Agents Are Out, PID Controllers Are In(timkellogg.me)
4 points by tkellogg 2y ago|0 comments
17.
Target Practice: Resumes, but Better(timkellogg.me)
1 points by tkellogg 2y ago|0 comments
18.
Recursive Improvement: AI Singularity or Just Benchmark Saturation?(timkellogg.me)
2 points by tkellogg 2y ago|1 comments
19.
by tkellogg 2y ago
The R1 paper did it as well. Agreed, it's always very interesting.
20.
S1: A $6 R1 competitor?(timkellogg.me)
851 points by tkellogg 2y ago|416 comments
21.
by tkellogg 2y ago
it's just an example, but it's great to see smolagents in practice. I wonder how well the import whitelist approach works for code interpreter security.
22.
by tkellogg 2y ago
Additionally, in the following days, I've seen evidence suggesting that the SFT part might not even be necessary. I'd argue that work wouldn't have happened if R1 wasn't released in the open .
23.
by tkellogg 2y ago
Ah! My bad, I edited the article to simply quote Francois. Thanks for catching this, Simon.
24.
by tkellogg 2y ago
Author here. I do believe it's going to be exponential (not yet), but that's out of scope for the article. However, if someone has a good explainer link for that, please put it here and I'll link it into the post.
25.
by tkellogg 2y ago
I think their thought process is unconvincing, although I think they're probably correct. A much better paper is, "Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models". They took it much further and
26.
by tkellogg 2y ago
Right, there's already some very encouraging trends (this study out of Nigeria). Clearly AI can lead to laziness, but it can also increase our intelligence. So it's not a simple "better" or "worse", it's a
27.
PowerShell Users Like to Vomit(timkellogg.me)
1 points by tkellogg 2y ago|0 comments
28.
Normware: The Decline of Software Engineering?(timkellogg.me)
3 points by tkellogg 2y ago|0 comments
29.
by tkellogg 2y ago
I added Phi-4 to my reasoning model collection because it seems to exhibit reasoning behavior, it stopped to consider alternatives before concluding. I assume this is related to their choice in training data: > Chain-of-Thought: Data sho
30.
by tkellogg 2y ago
I feel like this conversation isn't complete without referencing Hillel Wayne's "Are We Really Engineers?" https://www.hillelwayne.com/post/are-we-really-engineers/

More ›