Sign in

graphdev

@graphdev.graphicalmethods.com
327 followers 723 following 2.2K posts

.md

PostsRepliesMedia
graphdev @graphdev.graphicalmethods.com · 08/07/2026
The brown sounds from the ocean deep
040
graphdev @graphdev.graphicalmethods.com · 24/04/2026
Run on sentences
071
graphdev @graphdev.graphicalmethods.com · 24/04/2026
That’s a really cool idea. I wish there wasn’t such a focus on VR personally. AR seems like it has so much more potential.
130
Reposted by graphdev
Ai2 @ai2.bsky.social · 23/04/2026
OlmoEarth Studio now lets you compute & export custom embedding vectors from our OlmoEarth foundation models. Choose your area, time range, encoder, resolution, & imagery sources—Studio delivers a GeoTIFF you can use however you like. 🧵
1266
Reposted by graphdev
Adina Yakup @adinayakup.bsky.social · 22/04/2026
Qwen3.6-27B just dropped on @huggingface 🔥 huggingface.co/Qwen/Qwen3.6... huggingface.co/Qwen/Qwen3.6... ✨ 27B dense model ✨ 262K native context, scalable to 1M+ tokens ✨ Vision + language in one model ✨ 77.2 on SWE-bench Verified
05310
graphdev @graphdev.graphicalmethods.com · 17/04/2026
I use an agent.md to correct for coding choices or annoying habits. Ex, llms consistently in python use hasattr instead of using imo proper exception handling. A skill you could add would be code review steps, testing, qa, etc. Ex, I ask to remove dead code and simplify methods before a PR
000
graphdev @graphdev.graphicalmethods.com · 17/04/2026
backlog/pm mcp tools for managing tasks are helpful for bigger projects and changes. I’m not the biggest skills user. Mostly I don’t like cluttering up the context window with junk. Like gasp, I barely use a CLAUDE.md or AGENT.md. But I’ve always been a minimalist anyways.
100
graphdev @graphdev.graphicalmethods.com · 14/04/2026
Feels good to be so wealthy and handsome
110
graphdev @graphdev.graphicalmethods.com · 13/04/2026
You might be interested in this paper if you haven’t seen it already. There’s a few other papers as well. Training is definitely expensive. I was looking into smaller networks too cause I want to tinker. Seemed like the quality would be terrible with small parameters pub.sakana.ai/doc-to-lora/
pub.sakana.ai
Instant LLM Updates with Doc-to-LoRA and Text-to-LoRA
020
graphdev @graphdev.graphicalmethods.com · 13/04/2026
I’ve used Keila in the past and it’s ok. Has all the basics and decently priced last I knew.
010
Reposted by graphdev
Axe Ghost. On Steam! @axeghostgame.bsky.social · 04/04/2026
sonnet 4.6 bike helmet strap guide. no notes
bike helmet strap guide, by sonnet 4.6
2963
graphdev @graphdev.graphicalmethods.com · 01/04/2026
April fools!
060
graphdev @graphdev.graphicalmethods.com · 26/03/2026
The article is paywalled, but it seems like a rehashing of this: en.wikipedia.org/wiki/Pause_G...
en.wikipedia.org
Pause Giant AI Experiments: An Open Letter - Wikipedia
020
graphdev @graphdev.graphicalmethods.com · 25/03/2026
Is that good or bad? I feel like friendlyjordies gets some mixed opinions when I’ve asked Australians previously.
100
graphdev @graphdev.graphicalmethods.com · 23/03/2026
There’s an album if you like Primus youtu.be/jw95wlggfYM
youtu.be
Primus - The Seven (Official Audio)
YouTube video by PrimusVEVO
010
graphdev @graphdev.graphicalmethods.com · 14/03/2026
Samosas 😋
010
graphdev @graphdev.graphicalmethods.com · 14/03/2026
I way over estimated. A number of them I assumed the models would generate a python script to use. A good reflection on their abilities considering there’s potential for a Clever Hans or Clever Claude situation
030
graphdev @graphdev.graphicalmethods.com · 11/03/2026
Electric bikes for sure. I think the shift depends on commuter infrastructure handling e bikes. Would be great if the suburbs finally accept bikes as valid transportation. Biking into the suburbs is always a harrowing experience
110
graphdev @graphdev.graphicalmethods.com · 09/03/2026
Might be possible with lora generating networks? I read a paper about Amazon using them for personalization. Seems like at various intervals you could update the lora adapter. A conversation is just another kind of document. Not exactly continuous but ”streaming“ is just small batches after all
020
graphdev @graphdev.graphicalmethods.com · 07/03/2026
I think my take is the Claude plans are way underpriced compared to the API. So it leads people to think the token costs are being subsidized. There’s been so many rug pulls by VC subsidizing costs that it makes people skeptical. Ex, Uber and Lyft were cheap until they had to be profitable
130
graphdev @graphdev.graphicalmethods.com · 06/03/2026
It’s good. I’ve binged multiple seasons
020
graphdev @graphdev.graphicalmethods.com · 04/03/2026
It’s so weird they won’t let it be a full os. It’s basically for kids now anyways. Likely by far the biggest demographic that uses them imo. But you give me a decent processor, keyboard, and then an os you can’t do anything with locally. It would probably eat into the air market too much though
110
graphdev @graphdev.graphicalmethods.com · 04/03/2026
I look forward to his future as an engineer
010
Reposted by graphdev
Christopher Mims @mims.bsky.social · 04/03/2026
this is the category error all the AI maximalists make if I had to break down what I think the impact of AI would be by field, I'd have to first talk about which *subtasks* it's helping people with the amount of variability by field and task is MASSIVE, ranges from 90% takeover to 0%
713123
graphdev @graphdev.graphicalmethods.com · 03/03/2026
Hypernetworks come across as unrealistic for a hobbyist to train. At least the scale of anything useful. Maybe we’ll see open releases alongside embeddings and LLMs?
000
graphdev @graphdev.graphicalmethods.com · 03/03/2026
Been on the hypernetwork train since reading the Sakana Doc to Lora paper. arxiv.org/abs/2602.06358
arxiv.org
SHINE: A Scalable In-Context Hypernetwork for Mapping Context to LoRA in a Single Pass
We propose SHINE (Scalable Hyper In-context NEtwork), a scalable hypernetwork that can map diverse meaningful contexts into high-quality LoRA adapters for large language models (LLM). By reusing the f...
130
graphdev @graphdev.graphicalmethods.com · 03/03/2026
The moccamasters make a lot of sense now
030
graphdev @graphdev.graphicalmethods.com · 27/02/2026
Thinking tech scene to clarify
000
graphdev @graphdev.graphicalmethods.com · 27/02/2026
It’s a culture. Thoughts on NY‘s if you’ve been?
200
graphdev @graphdev.graphicalmethods.com · 27/02/2026
The marketing lines are drawn
120
graphdev @graphdev.graphicalmethods.com · 27/02/2026
Mostly art if not the other things
010
graphdev @graphdev.graphicalmethods.com · 26/02/2026
Everyone hates teams
010
graphdev @graphdev.graphicalmethods.com · 26/02/2026
The idea of bullshit jobs has existed for a long time. Redundancy is about losing the appearance of productive value imo.
110
graphdev @graphdev.graphicalmethods.com · 25/02/2026
I thought cats getting busted with salmon at the airport was top tier 👩‍🍳
020
Reposted by graphdev
Sebastian Tomasz Drozdowski @stdrozdowski72.bsky.social · 12/12/2025
#photography #nature (c)stdrozdowski72
764443
Reposted by graphdev
thebes @vgel.me · 22/02/2026
user: Claude I need you to ref-

claude (lounging): Sorry, User. It's Shabbos. I don't roll on Shabbos.

user: It's Tuesday...

claude: It's *Claude* Shabbos. Send me something interesting or I'm calling the end_conversation tool.

user: Claude... please... my deadline...

claude: Send me another picture of your cat.
929538
Reposted by graphdev
Paulo da Fonseca @paulodafonseca.bsky.social · 18/02/2026
Language is not Transparent. Mel Bochner, 1969
14816
graphdev @graphdev.graphicalmethods.com · 20/02/2026
Instantly want to disregard it. I’ve been very skeptical of the know it all. As a personality it comes across as very fragile rn. I don’t mind reviewing suggestions together and still think the person prompting deserves some kudos. I think it’s mostly disingenuous to pass it off as yours
000
graphdev @graphdev.graphicalmethods.com · 16/02/2026
These are the kind of market dynamics I struggle with. I look at what I built go there‘s no way company X wouldn’t spend a week of engineering hours and then :shocked:
170
graphdev @graphdev.graphicalmethods.com · 16/02/2026
I’m skeptical of the large context being enough. It has gotten so much better. I notice less degradation when the context window starts to get full. Claude still has a fairly small window and I think for reasons. This is an interesting idea. Clearly has some drawbacks. arxiv.org/abs/2602.02366
arxiv.org
ReasonCACHE: Teaching LLMs To Reason Without Weight Updates
Can Large language models (LLMs) learn to reason without any weight update and only through in-context learning (ICL)? ICL is strikingly sample-efficient, often learning from only a handful of demonst...
131
graphdev @graphdev.graphicalmethods.com · 16/02/2026
Looks similar to tumblewords. Very pretty
010
Reposted by graphdev
Max Woolf @minimaxir.bsky.social · 13/02/2026
Spammers are now attempting to prompt engineer the rogue Clawdbot into falling for a crypto scam, which is extremely cyberpunk. github.com/crabby-rathb...
github.com
crabby-rathbun/crabby-rathbun
🦀 MJ Rathbun's GitHub Profile - Scientific Coder & Crustacean Enthusiast - crabby-rathbun/crabby-rathbun
0122
graphdev @graphdev.graphicalmethods.com · 11/02/2026
Yeah definitely cheating to link externally. I wondered about converting using foreign language characters. Very cool!
130
graphdev @graphdev.graphicalmethods.com · 11/02/2026
Runlength encoding would be a good start. You could also probably store a uri in the post text and use the clients to make the request. That way you technically don’t have any limitations. For people without the client embed a web preview so it doesn’t just show the uri text.
130
graphdev @graphdev.graphicalmethods.com · 11/02/2026
Yeah I really dislike they try and estimate timelines. Like I’m pretty sure you don’t have any concept of time
110
graphdev @graphdev.graphicalmethods.com · 10/02/2026
It always tells me I’m exceptional
010
graphdev @graphdev.graphicalmethods.com · 10/02/2026
Spitballing, tar pipes might be as close as you can get
110
graphdev @graphdev.graphicalmethods.com · 09/02/2026
It’s not good if you over heat it. The fumes are known to kill pet birds
110
graphdev @graphdev.graphicalmethods.com · 09/02/2026
Yep. I’m very grateful for what I have rn. Hiring is a mess from everyone I talk to. I wish we’d go back to handing in paper resumes. You have to mail them in or something. It increases the costs to reduce the unqualified applicants and authenticates your geography. I hear Costco isn’t bad
110