Samhan
@samhan.bsky.social
📤 100
📥 446
📝 83
reposted by
Samhan
Gergely Orosz
24 days ago
If you work in tech, and are ever approached by any crypto sh*tcoin even for a harmless ask ("here's some money for you/your project! Take it! No strings attached, promise!"), a warning: Talked with someone who took it. Seemed too good to be true. Death threats to them and family followed.
6
105
15
reposted by
Samhan
Eugene Vinitsky 🍒
about 1 month ago
I have literally wanted this feature for years and
@mariaa.bsky.social
threw it together!
add a skeleton here at some point
4
53
4
reposted by
Samhan
Paul Frazee
about 1 month ago
If you enjoyed
@danabra.mov
's social filesystem blogpost, you might get a kick out of a literal exploration of the premise I tried using the dat p2p filesystem 6 years ago
www.youtube.com/watch?v=k0JV...
loading . . .
Beaker Next - Dev Diary #6 - Building a social feed application
YouTube video by Paul Frazee
https://www.youtube.com/watch?v=k0JVGXL-BQY
3
71
9
reposted by
Samhan
antirez
about 1 month ago
My bet is that AI policies will be soon avoided by most companies / projects in favor of quality centric policies.
7
28
2
reposted by
Samhan
Tudor Girba
about 1 month ago
Nice idea
add a skeleton here at some point
0
4
1
reposted by
Samhan
antirez
about 1 month ago
My latest video: the only SECRET!!111 you need to know to work well with coding agents.
www.youtube.com/watch?v=TJ6r...
loading . . .
Il trucco DECISIVO!!111 per lavorare coi coding agent
YouTube video by Salvatore Sanfilippo
https://www.youtube.com/watch?v=TJ6ruN-o0PA
0
7
1
reposted by
Samhan
antirez
about 2 months ago
Make sure to check
lobste.rs
comments on my blog post, if you are interested in the topic of AI coding. They are better than HN comments about the same post.
loading . . .
https://loste.rs
4
26
1
reposted by
Samhan
antirez
about 2 months ago
New blog post: Don't fall into the anti-AI hype.
antirez.com/news/158
21
275
80
reposted by
Samhan
dan
about 2 months ago
ok i've tried a few of these, and
@bobbby.online
's Deciduous (
notactuallytreyanastasio.github.io/deciduous/tu...
) is most intriguing so far it roughly matches what i want out of a tool like this not perfect but i recommend playing with it
add a skeleton here at some point
2
59
5
reposted by
Samhan
Nathan Lambert
about 2 months ago
I updated my list of reasoning models with solid tech reports for 2025! 2025-01-22 - DeepSeek R1 -
arxiv.org/abs/2501.12948
2025-01-22 - Kimi 1.5 -
arxiv.org/abs/2501.12599
2025-03-31 - Open-Reasoner-Zero -
arxiv.org/abs/2503.24290
2025-04-10 - Seed-Thinking 1.5 -
arxiv.org/abs/2504.13914
loading . . .
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning…
https://arxiv.org/abs/2501.12948
1
47
13
reposted by
Samhan
roon
5 months ago
might be time
44
255
18
reposted by
Samhan
Simon Willison
9 months ago
Some notes on the new DeepSeek-R1-0528 - a completely different model from the R1 they released in January, despite having a very similar name Terrible LLM naming has managed to infect the Chinese AI labs too
simonwillison.net/2025/May/31/...
loading . . .
deepseek-ai/DeepSeek-R1-0528
Sadly the trend for *terrible naming* of models has infested the Chinese AI labs as well. DeepSeek-R1-0528 is a brand new and much improved open weights reasoning model from DeepSeek, …
https://simonwillison.net/2025/May/31/deepseek-aideepseek-r1-0528/
2
67
6
reposted by
Samhan
Alexander Doria
about 1 year ago
In case it interests anyone, I managed to set up a demo of GRPO RL training in Colab. It’s an adaptation of Will Brown instant classic for math reasoning. Replace llama 1B with qwen 0.5b and inference with vllm. Full training in about 2 hours.
colab.research.google.com/drive/1bfhs1...
1
38
6
reposted by
Samhan
Chris Offner
about 1 year ago
“The drop in demand for IT professions could signal long-term shifts due to technological changes, particularly artificial intelligence.”
www.swissinfo.ch/eng/workplac...
add a skeleton here at some point
1
3
1
reposted by
Samhan
Pablo Rivera
over 1 year ago
Painting of a group of hatchling Tyrannosaurus rex- showcasing the fact that even these massive predators started out as fluffy little youngsters!
#paleoart
#dinosaurart
#trex
3
271
77
reposted by
Samhan
Dare Obasanjo
over 1 year ago
Reid Hoffman’s Financial Times op-ed captures Silicon Valley VCs’ hopes and fears about Trump. They hope for eased scrutiny on tech acquisitions, relaxed crypto/AI regulations, and nuclear energy support. Their fear? Elon Musk as a power broker picking winners and losers. Both outcomes seem likely.
8
105
19
reposted by
Samhan
Ben Recht
over 1 year ago
Many people are saying. ;)
hdsr.mitpress.mit.edu/pub/8dqgwqiu...
0
4
1
reposted by
Samhan
Marie Le Conte
over 1 year ago
I did it! I wrote up my Posting Middle Classes Theory, and tried to explain why I think Threads failed where Bluesky succeeded:
youngvulgarian.substack.com/p/down-in-th...
(free to read)
143
2183
579
reposted by
Samhan
Kenneth Marino
over 1 year ago
Daily paper: GPT-4V(ision) is a Generalist Web Agent, if Grounded
arxiv.org/abs/2401.01614
loading . . .
GPT-4V(ision) is a Generalist Web Agent, if Grounded
The recent development on large multimodal models (LMMs), especially GPT-4V(ision) and Gemini, has been quickly expanding the capability boundaries of multimodal models beyond traditional tasks like i...
https://arxiv.org/abs/2401.01614
1
4
2
reposted by
Samhan
Mark Riedl
over 1 year ago
Alibaba has their own version on GPT-o1. This might be the best description of “o1-type”systems so far
arxiv.org/abs/2411.14405
loading . . .
Marco-o1: Towards Open Reasoning Models for Open-Ended Solutions
Currently OpenAI o1 has sparked a surge of interest in the study of large reasoning models (LRM). Building on this momentum, Marco-o1 not only focuses on disciplines with standard answers, such as mat...
https://arxiv.org/abs/2411.14405
8
267
40
reposted by
Samhan
François Fleuret
over 1 year ago
Winter is here in Switzerland!
loading . . .
4
67
1
reposted by
Samhan
Annalee Newitz
over 1 year ago
If you're new to Bluesky, or just want to understand how it's profoundly different from platforms like X, I wrote this article last year about federated social media -- Bluesky and Mastodon specifically -- and how they fit into the US history of federalism.
www.theatlantic.com/technology/a...
loading . . .
Ben Franklin Would Have Loved Bluesky
Facebook and Twitter seem less relevant by the day. They may be replaced by new “federated” platforms.
https://www.theatlantic.com/technology/archive/2023/05/social-media-centralism-fediverse/674041/?gift=f1aD8da5ityMMOBvk08VCRgqbNhgRez118DFjXwTJAk&utm_source=copy-link&utm_medium=social&utm_campaign=share
29
702
174
reposted by
Samhan
Gergely Orosz
over 1 year ago
The #1 overall free app on the App Store (iOS) and Google Play (Android) is... Bluesky?? This is pretty incredible from an independent team, with a fraction of the budget as most other apps on these charts. (This represents current downloads, but it still huge. Congrats Bluesky team!)
4
256
13
reposted by
Samhan
Gergely Orosz
over 1 year ago
How cool is this: this is from the Bluesky firehose! What's so cool about this is: 1. The throwback style 2. That *any* dev can do this! Thanks to there being an API for this (no other social network allows this AFAIK) This API is a big reason to be bullish on Bluesky
firehose3d.theo.io
16
353
66
reposted by
Samhan
Kent Beck
over 1 year ago
Some folks look to be using AI as deflection—I don’t have to deal with my own fear/boundaries/control issues/blaming because AI will make programmers so efficient that there will be no conflicts to trigger my shit. Nope
2
29
5
reposted by
Samhan
Kent Beck
over 1 year ago
Untitled, 150 cm x 45 cm, acrylic on glass. These abstracts often start with a single gesture, in this case the sort of squid-looking figure on the left. The rest fills itself in around that.
1
34
2
reposted by
Samhan
antirez
over 1 year ago
In case it's useful for you as well, Word2Vec converted in a sane binary format.
huggingface.co/antirez/word...
loading . . .
antirez/word2vec-simple-parsing · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
https://huggingface.co/antirez/word2vec-simple-parsing
2
20
6
reposted by
Samhan
Gergely Orosz
over 1 year ago
Bluesky never felt so active as it has, the last few days. It's great to be here! (I'm here for software engineering and tech topics - both following these, and sharing more about them. I write The Pragmatic Engineer and host the podcast with the same name. E.g. a recent article)
13
436
23
reposted by
Samhan
Andy Baio
over 1 year ago
Nice,
@theo.io
made exactly what I was looking for. It works perfectly, found a ton of Bluesky people I didn't know were here!
loading . . .
Bluesky Network Analyzer
Find accounts that you don't follow (yet) but are followed by lots of accounts that you do follow.
https://bsky-follow-finder.theo.io/
43
399
191
reposted by
Samhan
Soumith Chintala
over 1 year ago
what are good starter packs for: AI researchers, AI Systems people, GenAI hackers, LLM enthusiasts?
7
44
2
reposted by
Samhan
Joe Beda
over 1 year ago
What do you call a sheep who can sing and dance? Lady Ba Ba. (It’s not her fault. She was shorn that way.)
3
12
1
Bluesky seems to be the first decentralised social network to truly take off. Just like Bitcoin protocol proved influential so will AT.
over 1 year ago
1
0
0
reposted by
Samhan
rich harris
over 1 year ago
if only 16yo me could have known how often adult me would need to solve some graphics programming problem with the ditty my maths teacher Mr Pickles taught us 🎵 twiddly dum, twiddly dee 🎶 around the moon is pi times d 🎵 but if a hole you want repaired 🎶 then you must use pi r squared
#mrpickles
7
131
4
reposted by
Samhan
Bluesky
over 1 year ago
it's official — 1,000,000 people have joined Bluesky in just the last day!!! welcome and thank you for being here 🥳
loading . . .
2299
86706
10331
reposted by
Samhan
Dileep George @dileeplearning
over 1 year ago
And here's a notebook derived from this to model the experiment from Charan's lab
@charan-neuro.bsky.social
in
www.nature.com/articles/s41...
Took me about 15 minutes.... Let me know which other experiment you want me to model next :-)
colab.research.google.com/drive/1TU6-l...
add a skeleton here at some point
0
1
2
reposted by
Samhan
Vicki
over 1 year ago
if unit tests keep the app from failing why don't they just make the entire app out of unit tests
9
112
14
reposted by
Samhan
rat king 🐀 👑
over 1 year ago
real disorienting to go from twitter — where i do a post to 200k followers and get five favs — over to bluesky where a post gets like 200 favs immediately
106
4188
152
reposted by
Samhan
David Pfau
over 1 year ago
One thing still missing here is good discussion of academic papers, so I guess I'll be the change I want to see in the world Really interesting results from Jacob Andreas's group showing great performance on ARC-AGI just from doing a few gradient descent steps at test time
arxiv.org/abs/2411.07279
loading . . .
The Surprising Effectiveness of Test-Time Training for Abstract Reasoning
Language models have shown impressive performance on tasks within their training distribution, but often struggle with novel problems requiring complex reasoning. We investigate the effectiveness of t...
https://arxiv.org/abs/2411.07279
7
183
25
reposted by
Samhan
Chris Paxton
over 1 year ago
Zeroth bot is a really cute open-source humanoid:
zeroth.bot
Creator Jingxiang Mo says its ~$350 bill of materials, build it yourself, with an NVIDIA isaac sim for policy learning and everything
loading . . .
0
5
3
reposted by
Samhan
Jaana Dogan ヤナ ドガン
over 1 year ago
We are rebuilding foundational layers of Google's model/agent inference stack, and we will soon hire some more people! If you are interested in pairing with me, please DM! If you are located either in the Bay Area, London, or Zurich, it'd help! Python and/or C++ expertise would be good to have!
3
53
9
reposted by
Samhan
Louis Gray
over 1 year ago
For those of you who still believe in science, it is well known that following the Bluesky users in this starter pack will make you more intelligent, successful and more attractive to your preferred gender. It will also help you win arguments and entertain you. Get started:
go.bsky.app/49C2jUo
add a skeleton here at some point
12
86
27
reposted by
Samhan
Julia Evans
over 1 year ago
folder gotchas (in the terminal)
wizardzines.com/comics/folde...
7
225
41
reposted by
Samhan
The Verge
over 1 year ago
Google’s AI ‘learning companion’ takes chatbot answers a step further
loading . . .
Google’s AI ‘learning companion’ takes chatbot answers a step further
Can an AI chatbot stimulate curiosity?
https://buff.ly/4etipP5
2
48
5
reposted by
Samhan
Kelsey Hightower
over 1 year ago
I'm studying and practicing like it's day one.
15
332
12
reposted by
Samhan
Dileep George @dileeplearning
over 1 year ago
ICYMI...if you are curious about how the ARC-AGI challenge is connected to the idea of 'cognitive programs', perceptual+cognitive representations etc. have a read:
blog.dileeplearning.com/p/how-to-get...
loading . . .
How to get a million bucks: quick thoughts on cognitive programs and the ARC challenge.
The abstract reasoning corpus (ARC) challenge by Francois Chollet has gained renewed attention due to the 1M prize announcement. This challenge is interesting to me because the idea of “abstraction” a...
https://blog.dileeplearning.com/p/how-to-get-a-million-bucks-quick
2
15
3
reposted by
Samhan
Brian Tyler Cohen
over 1 year ago
It really is amazing that we spent more than a year begging Meta to simply allow political content to be seen on Threads, and yet they’re so hostile to their own audience that Bluesky is looking like it’s going to lap Threads.
523
13093
1165
reposted by
Samhan
mr potato
over 1 year ago
cashier: can take your order? me: [spent the entire time in line rehearsing for this moment] yes, one pleaseburger cheese :)
123
10959
1872
reposted by
Samhan
Alexandria Ocasio-Cortez
over 1 year ago
I don’t even know why I stopped using this in the first place? Good GOD it’s nice to be in a digital space with other real human beings. Can someone give a Cliff Notes of what went down here over the last year? Tea, dev notes, big memes, whatever is best please
2669
57522
3384
Load more
feeds!
log in