Viewing profile — ErikBjare
ErikBjare
HN member- Joined
- Wed, Mar 26, 2014, 7:56 AM UTC
- HN karma
- 335
- Public activity
- 365 items
- HN profile
- View on Hacker News ↗
About ErikBjare
Socials:
- github.com/ErikBjare
Interests:
AI/ML, Open Source, Startups
---
# Building
gptme, powerful AI assistant/agent in your terminal: https://github.com/gptme/gptme
ActivityWatch, the "world's best" free and open-source time tracker: https://activitywatch.net
# Links
Website: https://erik.bjareholt.com/
GitHub: https://github.com/ErikBjare
Recent public activity
-
comment
Comment #49107580
I don't think I've ever had a keyboard die on me
-
comment
Comment #48561646
I refuse to believe this wasn't written to intentionally bait, reads like copypasta.
-
comment
Comment #47370772
Both of those are statically served
-
comment
Comment #47146277
If you run Claude Code with `-p --output-format json` it's no longer a TUI, but it's still a harness.
-
comment
Comment #47123004
The subsidized Codex/Claude subscriptions make it not so bad.
-
comment
Comment #46753733
To be fair to Android, iOS isn't offering "good device independent camera apps" either, you only have ~one choice of device with iOS.
-
comment
Comment #46718395
An informative rejection message with the reason for the restriction usually addresses this well with recent models.
-
comment
Comment #46713123
Those requirements should include "estimated setup & recurring administrative costs for businesses of size x to comply should be lower than y" in some way.
-
comment
Comment #46652114
And mine last year
-
comment
Comment #46635428
https://erik.bjareholt.com
-
comment
Comment #46623795
That's where it got its name from
-
comment
Comment #46368161
They were assigned a huge prefix at the creation of the internet iirc
-
comment
Comment #46348126
Plenty of aider-era tools were though, like my own gptme which is about as old as aider
-
comment
Comment #46310368
This was confusing me, thanks.
-
comment
Comment #46044012
Kinda disappointed, doesn't seem all that advanced to me.
-
comment
Comment #45956423
I did that a year ago, I imagine it would work better today.
-
comment
Comment #45879970
Same with mine that has been around for over 2 years now: https://github.com/gptme/gptme
-
comment
Comment #45876163
In the case of BBs they need to withstand a fair bit of force, so I don't think it's about money. The biodegradable ones (PLA) on the market I've tried perform less well than their…
-
comment
Comment #45865804
AFAIK (not much) it definitely helps to train on longer sequences even with rope/yarn and is needed if you care about long context performance (and not just the long context capabi…
-
comment
Comment #45865743
Apparently Persona was even based on some prior work called "VerifiedEmailProtocol", eerily similar to the OP
-
comment
Comment #45758286
This summarizes it well. I wish Codex could learn to tooluse as well as Sonnet, would really unleash the deep thinking it's so good at.
-
comment
Comment #45757004
Honestly I think poetry was a bigger development than uv. I used pipenv before it, and requirements before that, and I can't imagine going back. I've yet to fully embrace uv and mi…
-
comment
Comment #45626248
> I am as much in the dark as anyone on if you can get LLMs to reliably do useful work. I can't understand how people still say this with a straight face.
-
comment
Comment #45577956
I've been building a Python+Qt+PyInstaller app for a decade (ActivityWatch) which we are migrating to Tauri for a wide range of reasons. Perhaps the largest one is easy cross-platf…
-
comment
Comment #45577914
If you generate the tests, run those tests, apply fixes without any oversight, it is the very same situation. In reality, we have PR reviews.