Sign in

Patrick Ohlson

@patrick-ohlson.ms-dot.net
1.6K followers 477 following 2.2K posts

404 YT: youtube.com/@paomakes Signal : @pao.1337

PostsRepliesMedia
Patrick Ohlson @patrick-ohlson.ms-dot.net · 4m
Spoooky🔥❤
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 4m
I think this would trigger a debate between @why.bsky.world and @jcsalterego.bsky.social So, I'm fetching popcorn
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 33m
JIT-memory coding continues Astra is about to verify against original paper not started llm training yet
Screenshot of JITM Admin
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 1h
There can be non slop in AI It still require that a human learn(ed) things though This goes for non developer things also Anyways, just old man shouting at clouds
Agreed. We should use one small, verifiable milestone per sprint, then pause for a paper-alignment check before choosing the next.
Each sprint will record:
- Purpose: which paper mechanism or practical requirement it advances.
- Acceptance: observable results needed to finish.
- Verification: relevant tests and live-use evidence.
- Paper check: preservation, navigation, query-time research, sufficiency, downstream use, and deviations.
- Closeout: updated handoff/priorities, verified memory note, and focused commit.
Admin improvements are supporting work; they won’t count as progress toward missing paper mechanisms.
Next: finish the current private-browser/metrics sprint, then prioritize source preservation and a bounded retrieval-quality baseline.
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 2h
Ok, won't lie That 1340km drive yesterday... Is now starting to feel... everywhere 🤣
020
Patrick Ohlson @patrick-ohlson.ms-dot.net · 4h
@coolhand.bsky.social 😿🤬
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 5h
These things are why I don't mind our somewhat high taxes(Sweden)
020
Patrick Ohlson @patrick-ohlson.ms-dot.net · 9h
S*d off with these f*c'n apps, brain get super envy Cnt 😉🤗❤️ Oh, I'm back
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 9h
❤️ It is getting colder, but still warmer than usual 🤷‍♀️ A bit off today, the 1340km ride yesterday 🤣
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 20h
Back home That's it, that's the skeet
110
Patrick Ohlson @patrick-ohlson.ms-dot.net · 07/10/2026
New sol seem to be a token eater also, but have same plan It's been Astra-> Sol -> Luna now
010
Reposted by Patrick Ohlson
Patrick Ohlson @patrick-ohlson.ms-dot.net · 06/10/2026
Cleaned up some artifacts But live dev is still on track From a camper somewhere in the woods outside of Berlin
Codex screenshot
Showing memory mcp logs
111
Reposted by Patrick Ohlson
Unsloth AI @unsloth.ai · 06/10/2026
Google releases EmbeddingGemma 2, a new open model that runs locally on 0.5GB RAM. The 740M parameter embedding model combines a 270M text model with vision (170M) + audio (300M). Run & train the model via Unsloth. GGUF: huggingface.co/unsloth/embe... Guide: unsloth.ai/docs/models/...
3989
Patrick Ohlson @patrick-ohlson.ms-dot.net · 06/10/2026
Cleaned up some artifacts But live dev is still on track From a camper somewhere in the woods outside of Berlin
Codex screenshot
Showing memory mcp logs
111
Patrick Ohlson @patrick-ohlson.ms-dot.net · 06/10/2026
Most I've seen is on tabloid level But this was not, but I agree
110
Patrick Ohlson @patrick-ohlson.ms-dot.net · 06/10/2026
Interesting read indeed, for once
100
Patrick Ohlson @patrick-ohlson.ms-dot.net · 06/10/2026
Ah, was at a beergarden, but yes This looks like another nice addition🔥🔥
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 06/10/2026
Oooh, the orange one🎉🔥
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 06/10/2026
Ok, first day with ChatGpt 100 Using Astra(Light) as orchestrator for Sol/Luna Landed at 43% left👀 Conclusion: Continue using Claude 200 for grunt work Use Astra for the creative stuff
000
Reposted by Patrick Ohlson
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Memories, not lost like tears in the rain👀
Screenshot of agent markup with memory links
011
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
No WIndows XP?😉
110
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Memories, not lost like tears in the rain👀
Screenshot of agent markup with memory links
011
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Codex restart, memory mcp initiated quest continues
The relaunch worked: this chat now exposes all 18 JIT memory tools. Both session resume and budget inspection succeeded through the registered MCP. The session is active at checkpoint 11, with 98/100 requests used. I’ll record this host-verification milestone without spending retrieval requests.
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Jerry, no
190
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
This is awesome I so wanted a zx, coz my first was a z80 based But then I landed on Amiga Still wonder about outcome of a Sinclair journey at times
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
First change was letting Astra control sub agents, then move to the full uv/ruff/lint world Also tidied up admin for memory
Screenshot of basic agent memory admin
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Nah, it was the use of Fable and delegate to sub models for cost efficiency Sorry, squirrel brain🤣🤣
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Or did you mean the agent thing?🤷‍♀️🎉😉
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Codex? Or what's it doing? I saw this 👇, so I started a Sunday live implemention thing when I tested Astra Using the session to remember and build at the same time
arxiv.org
Just-In-Time Agent Memory with Runtime Agentic Research
200
Reposted by Patrick Ohlson
Dr Lucy Rogers @drlucyrogers.com · 05/10/2026
Need a senior engineer in #Aotearoa New Zealand - but not enough to justify sending someone over? I’m available for short engineering assignments. I’m good at practical eng., unusual environs, prototypes, mfg & troubleshooting. Need a trusted pair of technical eyes & ears? Contact me!
Woman in a gold jacket and a drill in her hand.
04028
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Adjusted your agent idea to Astra, @timkellogg.me It has served me well with Fable, so ty again
100
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Context
Screenshot of Codex with Astra light, delegating to subagents
100
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Have this one, had !!, if old man recalls right🤔
Typewriter Olympia Model 8 , Swedish, circa 1935.
Pictures
100
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Ye, most likely👍 But I consider this a training run Claude max20 will offload also
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Got an idea Let Codex Astra(light) use the gpu manager tool for rent of a training instance in this live jitm session Might be a epic failure, or not🤔🤷‍♀️🥳
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Dear reader He did upgrade to pro 100 We shall see the outcome of this decision
120
Patrick Ohlson @patrick-ohlson.ms-dot.net · 05/10/2026
Chernobyl was kinda scary Fish and blueberries got hit hard here
210
Reposted by Patrick Ohlson
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Ending for the night Well done by Astra on the chatgpt plus account Had to use 3 resets tho😂
The multi-worker MCP exercise passed:
- Two jobs completed, including abandoned-lease recovery.
- Stale commits, revoked access and purged-evidence replay were rejected.
- Used 9/9 budget units across 44 total calls.
- Recovered 28 events after reconnect.
75 regression tests passed. These were scripted workers, not LLM agents.
[Handoff updated](D:/Dropbox/Git/Codex/JIT-Agent/docs/HANDOFF.md). Next: job discovery and safe controls for closed or expired sessions.
131
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Oh yes
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Being male and aphantasic, this is😿😉🤗
110
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Niceee
github.com
GitHub - Patrick-Ohlson/AmbrogioMower: Ambrogio L200 Custom firmware
Ambrogio L200 Custom firmware. Contribute to Patrick-Ohlson/AmbrogioMower development by creating an account on GitHub.
100
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
I've used gcc for h8s2144, sh3 should work?
100
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Oh, saw one here in Sweden now Might buy(300lx) See that it's Hitachi/renesas sh Are there any current repos?
110
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
s/bee/be Lex Editbutton, @pfrazee.com
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Addendum There must always bee room for Jerry, no
110
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
that's alot of yes on a Sunday, even for you Jerry
130
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Going to test if Claude Fable can take over the ball tomorrow👀
000
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Ending for the night Well done by Astra on the chatgpt plus account Had to use 3 resets tho😂
The multi-worker MCP exercise passed:
- Two jobs completed, including abandoned-lease recovery.
- Stale commits, revoked access and purged-evidence replay were rejected.
- Used 9/9 budget units across 44 total calls.
- Recovered 28 events after reconnect.
75 regression tests passed. These were scripted workers, not LLM agents.
[Handoff updated](D:/Dropbox/Git/Codex/JIT-Agent/docs/HANDOFF.md). Next: job discovery and safe controls for closed or expired sessions.
131
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Tested another thing, will continue tomorrow But, $2.67 is better than $5.07/hr for unrestricted GLM 5.3 flash
V100 box (1Cat fork)	Czech box (Studio)
Decode	38.6 tok/s	32 tok/s
Prefill, 12.5k prompt	299 tok/s	519 tok/s
Context	64k, with room for ~7 full requests	64k, one slot
Price	$2.67/hr	$5.07/hr
010
Patrick Ohlson @patrick-ohlson.ms-dot.net · 04/10/2026
Jaz, it's Sunday😉🤗🔥
050