kovek

6 ms·

1.
by kovek 5d ago
Would the loss at the global maxima be much lower that the loss at most local maxima?
2.
by kovek 1mo ago
I think to learn a language you need to do as many modalities as you can: Speaking, listening, reading, singing, asking questions to native speakers, asking questions to LLMs, writing, flash cards, etc. The obstacle with LLMs is that they a
3.
by kovek 2mo ago
Try being an English speaker learning Mandarin from Claude
4.
by kovek 2mo ago
I think learning from another human being live who can understand your confusingly articulated questions and who can adjust their answers for you, another human being, that's great.
5.
by kovek 2mo ago
Please support /voice for claude!
6.
by kovek 2mo ago
I think it's easy to lie to consumers. Then, how much agency do they have?
7.
by kovek 3mo ago
I have only one thing to say.... what
8.
by kovek 3mo ago
I heard people say this before. I'm wondering, how do you instruct the LLM to generate the tests? Do you tell it the scenarios that would be covered, or do you just tell it to write tests for the code?
9.
by kovek 3mo ago
I thought it was known since a few years now that if you train models to NOT do certain things, then they start behaving in weird ways…
10.
by kovek 4mo ago
What if the organization who tries to verify sends a request on an app on the user’s iPhone (or whatever device can do the same), and the user scans their face with FaceID to produce a file send to the organization, which will then send tha
11.
by kovek 4mo ago
I’d say it’s possible to have creativity when you’re sitting as well. I like to think that’s it’s all about staying active. Reading, diarying, calling a friendind. All of that.
12.
by kovek 4mo ago
LLMs are like a search engine that autocompletes. It's a tool.
13.
by kovek 4mo ago
What negative consequences does being unelected have?
14.
by kovek 4mo ago
Maybe you can't 100% know what every layer "thinks", if you go through all the layers, you might see a cohesive "thinking" story. So, if there is any information you lose at layer N, you might learn some of it in la
15.
by kovek 4mo ago
I’ve read recently about natural systems in the book Antifragile. It’s interesting how those systems can become better.
16.
by kovek 4mo ago
> The core idea is content-dependent selection. For each query, the model selects which parts of the sequence are worth attending to, and computes attention exactly over those positions. I don't know if this will help for things lik
17.
by kovek 5mo ago
I don’t think triaging is necessarily an easy task
18.
by kovek 5mo ago
Does thinking about how to offload matter?
19.
by kovek 5mo ago
10s of GBs? ( 1,000,000 context * 1,000 vector size ) ^ 2 = 1,000,000,000,000,000,000… oh wow.. I must be miscalculating What about only storing the conversation and then recomputing the embeddings in the cache? Does that cost a lot? Doing
20.
by kovek 5mo ago
What if the cache was backed up to cold storage? Instead of having to recompute everything.
21.
by kovek 6mo ago
Is this similar to send 48656c6c6f2c20686f772061726520796f753f in the prompt? As done here: https://youtu.be/GiaNp0u_swU?si=m7-LZ7EYxJCw0k1-
22.
by kovek 6mo ago
> You seem to be hinting at the "chemical imbalance" theory of antidepressants, which has been largely debunked Can you say more?
23.
by kovek 6mo ago
Models need pre-training and fine tuning. Humans can do online learning.
24.
by kovek 7mo ago
What if we asked users if they want extra protection? I think that would be nice..
25.
by kovek 7mo ago
For the tech docs writing, just give me the bullet points and I'll send them to the AI and discuss the bullet points with it.
26.
by kovek 7mo ago
I was thinking about the ability of representing different kinds of numbers. Imagine that we had a certain CPU that could process algorithms, and the final output of the algorithm is a number. The CPU has a certain number of operations (At
27.
by kovek 7mo ago
What is there to be furious about?
28.
by kovek 7mo ago
Every word and every hierarchy of words in natural language is understand by LLMs as embeddings (vectors). Each vector has many many dimensions, and when we train the LLMs, their internal understanding of those vectors sees all sorts of dim
29.
by kovek 7mo ago
I think that semantically this question is too similar to the car wash one. Changing subjects from car to elephant and car wash to creek does not change the fact that they are subjects. The embeddings will be similar in that dimension.
30.
by kovek 7mo ago
Does https://happy.engineering/ need to use the API keys or can use oauth? It's basically a frontend for claude-cli.

More ›