graphdev's Avatar

graphdev

@graphdev.graphicalmethods.com

.md

304
Followers
694
Following
2,149
Posts
29.10.2024
Joined
Posts Following

Latest posts by graphdev @graphdev.graphicalmethods.com

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

07.03.2026 16:15 πŸ‘ 3 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

It’s good. I’ve binged multiple seasons

06.03.2026 00:24 πŸ‘ 2 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

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

04.03.2026 23:49 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

I look forward to his future as an engineer

04.03.2026 21:47 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

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%

04.03.2026 16:55 πŸ‘ 133 πŸ” 23 πŸ’¬ 7 πŸ“Œ 2

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?

03.03.2026 17:20 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
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...

Been on the hypernetwork train since reading the Sakana Doc to Lora paper.

arxiv.org/abs/2602.06358

03.03.2026 17:20 πŸ‘ 4 πŸ” 1 πŸ’¬ 1 πŸ“Œ 0

The moccamasters make a lot of sense now

03.03.2026 06:52 πŸ‘ 4 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Thinking tech scene to clarify

27.02.2026 18:35 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

It’s a culture. Thoughts on NYβ€˜s if you’ve been?

27.02.2026 18:33 πŸ‘ 0 πŸ” 0 πŸ’¬ 2 πŸ“Œ 0

The marketing lines are drawn

27.02.2026 16:18 πŸ‘ 2 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Mostly art if not the other things

27.02.2026 13:29 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Everyone hates teams

26.02.2026 17:13 πŸ‘ 2 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

The idea of bullshit jobs has existed for a long time. Redundancy is about losing the appearance of productive value imo.

26.02.2026 15:49 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

I thought cats getting busted with salmon at the airport was top tier πŸ‘©β€πŸ³

25.02.2026 22:03 πŸ‘ 2 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Post image

#photography #nature (c)stdrozdowski72

12.12.2025 18:19 πŸ‘ 622 πŸ” 41 πŸ’¬ 7 πŸ“Œ 0
22.02.2026 18:25 πŸ‘ 288 πŸ” 39 πŸ’¬ 9 πŸ“Œ 3
Post image

Language is not Transparent. Mel Bochner, 1969

18.02.2026 15:36 πŸ‘ 45 πŸ” 15 πŸ’¬ 1 πŸ“Œ 1

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

20.02.2026 09:26 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

These are the kind of market dynamics I struggle with. I look at what I built go thereβ€˜s no way company X wouldn’t spend a week of engineering hours and then :shocked:

16.02.2026 19:45 πŸ‘ 7 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
Preview
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...

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

16.02.2026 14:14 πŸ‘ 3 πŸ” 1 πŸ’¬ 1 πŸ“Œ 0

Looks similar to tumblewords. Very pretty

16.02.2026 02:19 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
crabby-rathbun/crabby-rathbun πŸ¦€ MJ Rathbun's GitHub Profile - Scientific Coder & Crustacean Enthusiast - crabby-rathbun/crabby-rathbun

Spammers are now attempting to prompt engineer the rogue Clawdbot into falling for a crypto scam, which is extremely cyberpunk. github.com/crabby-rathb...

13.02.2026 04:19 πŸ‘ 13 πŸ” 2 πŸ’¬ 0 πŸ“Œ 0

Yeah definitely cheating to link externally. I wondered about converting using foreign language characters. Very cool!

11.02.2026 17:57 πŸ‘ 3 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

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.

11.02.2026 15:33 πŸ‘ 3 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Yeah I really dislike they try and estimate timelines. Like I’m pretty sure you don’t have any concept of time

11.02.2026 03:53 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

It always tells me I’m exceptional

10.02.2026 16:36 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Spitballing, tar pipes might be as close as you can get

10.02.2026 05:12 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

It’s not good if you over heat it. The fumes are known to kill pet birds

09.02.2026 05:54 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

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

09.02.2026 01:39 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0