kagen (@kagnar_) header
kagen (@kagnar_) profile picture

doing 0 to 1 @getodinai @nerobotai

Joined April 2011
Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)Photo by kagen (@kagnar_)
I got married last month, and spent the furthest time away from work during that period and the honeymoon Unplugging while we broke ATHs really helped me reflect on what I’ve learned up until this milestone and what’s important to me in life I wanted to share this publicly mostly to document my thoughts. I hope it’s helpful but it’s not intended to be life advice or anything like that. Just my own musings which include some stories, quotes, and pictures here and there.
134
131
29
1,138
238,161
kagen retweeted
Open call. I’ll pay someone 400$ to make instructor support Claude function calling and that xml mess. Lmfao. With patch(create=) Just needs a PR with passing tests. Negotiable.
10
4
1
57
9,460
Someone wants to make this work, with a small DSPy program and some optimizer runs?
Function calling accuracy in LLMs really sucks. The best function calling accuracy is obtained with GPT-4 and it's 83.8%. It's already too low to be practical, but one should discount this number more assuming that the test data Berkeley used to evaluate function calling contained up to 4 functions to choose from. If with <=4 functions to choose from the accuracy is ~80%, imagine how low it would be when we will have 20 functions to choose from. What if we want to call 200 functions? And this leaderboard only measures the function choice accuracy. It doesn't consider the choice of parameters and their values. In some papers I read, the accuracy of function calling is estimated to be near 50%. And with these numbers, they talk about LLM-based agents. Agents, Carl! gorilla.cs.berkeley.edu/lead…
3
9
55
11,734
I always thought these remarks are not very serious but since they’re common it could be instructive to comment. Tradeoffs. Context (concatenation) & Retrieval pose fundamentally different tradeoffs in recall, precision, and latency/cost. You can’t “choose” between RAM and disk.
With the context size of 10M tokens, de we still need vector databases?
15
12
3
130
25,306
Here's one of proof in the tech report, Gemini team compared Gemini 1.0 pro w/ RAG v.s. Gemini 1.5 Pro. Gemini 1.5 outperform Gemini 1.0 + RAG RAG technics they used, TF-IDF database + Cosine similarity search + Rerank + Similar passages Note: TF-IDF database - frequency-inverse document frequency and it is a measure, used in the fields of information retrieval (IR) and machine learning
6
7
2
52
16,059
Nero is seriously one of the most underrated AI products out there. @kagnar_ is way too busy building to advertise. We have an integration on @AgentHub_AI as well if you want a simple interface for it – at San Francisco, CA
Quick demo of @nerobotai AI Crawl Assistant in action. #buildinpublic It's able to navigate sitemaps for desired pages all from natural language inputs. It does so by leveraging a set of tools and/or code interpreter to handle as many types of website structures as possible. Currently we are using @OpenAI Assistant API and Mistral models. The beautiful frontend was made possible by the boilerplate framework @makerkit_dev from @gc_psk And shoutout to @KillerTopic who did a lot of the heavy lifting on the backend.
1
3
5
1,076
Every single one of these videos is AI-generated, and if this doesn't concern you at least a little bit, nothing will The newest model: openai.com/sora (Remember Will Smith eating spaghetti? I have so many questions)
1,691
7,662
3,125
55,306
9,637,105
RESPOND WITH JSON ONLY. DO NOT UNDER ANY CIRCUMSTANCES RETURN ANY CONVERSATIONAL TEXT.
46
188
32
3,545
329,273
🚨RAG is dead, long live RAWDOG🚨 Recursive Augmentation With Deterministic Output Generations
17
27
8
570
97,922
David Attenborough is now narrating my life Here's a GPT-4-vision + @elevenlabs python script so you can star in your own Planet Earth:
679
4,358
1,282
25,525
3,997,627
8
7
3
113
12,724
“I don’t think there has been any greater engine of inequality than the Federal Reserve Bank of the United States the last 11 years.” - Stanley Druckenmiller Exactly as bitcoiners have been saying. Eventually everyone will understand. xitter.cf/NorthmanTrader/status/…
98
831
155
3,141
Twitter on this fine Monday morning
21
28
4
641
Buys on Coinbase are not completing. Coinbase is $350 lower than other exchanges right now, it's throwing off derivative indexes and likely impacting trade algos.
We’re investigating an issue impacting transactions on Coinbase.com and the mobile apps. The record of a recently initiated transaction may be delayed in showing up in your Coinbase account. There may also be some issues with some buys completing on the platform.
101
174
62
1,071
“There is no art which one government sooner learns of another than that of draining money from the pockets of the people.” —Adam Smith
2
53
1
154
SERVICE UPDATE We are experiencing DDOS attacks that can make it difficult to reach our servers. We are blocking the attacks, but it may happen that the exchange (website and API) is temporarily difficult to reach.
2
4
3
29
Bitcoiners! The War for Sound Money is upon us! Time to get locked & loaded!👇
1/ The Bitcoin Arsenal is a community resource brought to you by Swan Bitcoin and a host of contributors. More than just memes, the Bitcoin Arsenal is a curated compendium of tools for educating precoiners, newcoiners, and altcoiners. swanbitcoin.com/arsenal
3
3
1
17
Glad that innovative companies like @Tantra_Labs are leveraging the native security of #bitcoin multisignature to ensure that their client BTC is safe. Psyched to be collaborating with @block26vc, @Willwaukee, @Russell_LaCour, @kagnar_, @BVBTC and the rest of the Tantra team
1
6
18
The 2020s are the Remote Work decade A few predictions of what is likely to emerge [ a thread ] 💻🏠🌍
101
973
500
3,451