bguberfain

8 ms·

1.
by bguberfain 11d ago
If you are planing to watch the eclipse, a good resource is to look at http://xjubier.free.fr/en/site_pages/SolarEclipsesGoogleEart... In 2028 it will go straight to Sydney, Australia
2.
by bguberfain 11d ago
Nowadays we use LLMs mostly for doing agentic-based work. LLMs new Pareto frontier only make the headlines if they push the boundaries on benchmarks that are deterministic tasks. So models are encouraged to focus on these deterministic task
3.
by bguberfain 4mo ago
It is good to se big companies like Microsoft launching LLMs. They have large amount of compute power and good scientists to create useful models.
4.
by bguberfain 4mo ago
Any plans to port to sglang or vLLM?
5.
by bguberfain 4mo ago
It seems to be something related the moving average calculation. So it is just a glitch on the chart.
6.
by bguberfain 5mo ago
This guy seems to be talking seriously.
7.
by bguberfain 5mo ago
Not to demerit the recording, but I felt more nostalgic for the last sentence of the article "Sometimes, the internet is good" than for the musics itself.
8.
by bguberfain 5mo ago
We all know it... but I think they were very bold in this warning about using your private messages to train public models. _Your messages with AIs will be used to improve AI at Meta. Don't share information, including sensitive topics
9.
by bguberfain 6mo ago
"A watchdog kernel thread monitors RAM and NVMe pressure and signals userspace before things get dangerous." - which kind of danger this type of solution can have?
10.
by bguberfain 9mo ago
We can finally search for playlists with a giving song! A basic feature that Spotify is missing!
11.
Ask HN: Is it possible to implement this button on a browser?
1 points by bguberfain 10mo ago|1 comments
12.
by bguberfain 1y ago
So they used a LLM with knowledge cut in mid 2023 to evaluate 2023? Seems like a classic leakage problem. From paper: "testing set: January 1, 2023, to December 31, 2023" From the Llama 2 doc: "(...) some tuning data is more
13.
by bguberfain 1y ago
I think that there may be another solution for this, that is the LLM write a valid code that calls the MCP's as functions. See it like a Python script, where each MCP is mapped to a function. A simple example: def process(param1, p
14.
Google bug blocked Miniforge on GitHub(github.com)
3 points by bguberfain 1y ago|0 comments
15.
by bguberfain 1y ago
https://blogs.windows.com/windowsdeveloper/2025/05/19/the-wi...
16.
by bguberfain 1y ago
Not available in my country :(
17.
by bguberfain 1y ago
Unfortunately, it uses Miniconda, which does not allow usage in companies with more than 200 employees. I think it conflicts with AGPL license. I created a PR to fix that.
18.
OpenAI contest on X is void in Brazil, Italy, Quebec and others(openai.fm)
2 points by bguberfain 1y ago|0 comments
19.
by bguberfain 2y ago
Can you provide more information about this “bigger teacher” model?
20.
by bguberfain 2y ago
Until GPT-4.5, GPT-4 32K was certainly the most heavy model available at OpenAI. I can imagine the dilemma between to keep it running or stop it to free GPU for training new models. This time, OpenAI was clear whether to continue serving it
21.
by bguberfain 2y ago
Any chance you could release the dataset to the public? I imagine NewsCatcher and Polymarket might not agree..
22.
by bguberfain 2y ago
It remembers me Theano [0]. [0] https://en.wikipedia.org/wiki/Theano_(software)
23.
by bguberfain 2y ago
I agree. One image of what it is doing would improve the comprehension of the algorithm.
24.
by bguberfain 2y ago
The file I mentioned is just the begining... there is a folder full of .dll files, renamed to .pyd. I understand that this is the proprietary part, that limits usage for 30 minutes, but I think it is too closed for a MIT license.
25.
by bguberfain 2y ago
Thanks for sharing this! But I have some doubts about hidden installation procedures. It imports all functions from one_click (from one_click import *), which points to a compiled file. It then runs functions like install_webui and install_
26.
Alibaba Cloud fights back Anaconda(dockets.justia.com)
1 points by bguberfain 2y ago|0 comments
27.
by bguberfain 2y ago
"Nemotron-4-340B-Instruct is a chat model intended for use for the English language" - frustrating
28.
by bguberfain 2y ago
From the code, it seems to send information to https://vizly-notebook-server.onrender.com/ when in "production". Not so local (src: https://github.com/squaredtechnologies/thread/blob&#x2F
29.
OpenAI demo some ChatGPT and GPT-4 updates on 13 May(openai.com)
46 points by bguberfain 2y ago|30 comments
30.
Rerank 3: A new foundation model for efficient enterprise search and retrieval(txt.cohere.com)
45 points by bguberfain 2y ago|5 comments

More ›