Kevin Jacobs

43.9K posts

Kevin Jacobs banner
Kevin Jacobs

Kevin Jacobs

@kevinjaybe

Full Stack Dev & jack-of-all-trades. Fmr. analyst at Euroclear Bank (FundSettle) with 8y experience in funds. Biotech, news & tech addict.

Ghent, Belgium Katılım Ağustos 2010
142 Takip Edilen884 Takipçiler
Kevin Jacobs
Kevin Jacobs@kevinjaybe·
These LLM benchmarks really starting to look like the equivalent to CPU or GPU benchmarks. They say something but not everything about real life performance. If a model is within 2% this is negligible in day to day performance.
English
0
0
0
15
Kevin Jacobs
Kevin Jacobs@kevinjaybe·
Absolutely retarded to consider this anything but an enormous clusterfuck and loss.
Karoline Leavitt@PressSec

This is a victory for the United States that President Trump and our incredible military made happen. From the very beginning of Operation Epic Fury, President Trump estimated this would be a 4-6 week operation. Thanks to the unbelievable capabilities of our warriors, we have achieved and exceeded our core military objectives in 38 days. More on that tomorrow morning from @SecWar and Chairman Caine! The success of our military created maximum leverage, allowing President Trump and the team to engage in tough negotiations that have now created an opening for a diplomatic solution and long-term peace. Additionally, President Trump got the Strait of Hormuz reopened. Never underestimate President Trump’s ability to successfully advance America’s interests and broker peace.

English
0
0
0
35
Kevin Jacobs
Kevin Jacobs@kevinjaybe·
This seems to be a horrendous agreement. Nothing art of the deal about it. Trump got called and blinked. He is a terrible sales guy and an even worse president. If you still can’t admit this, you need to seek mental help.
English
0
0
0
29
Kevin Jacobs
Kevin Jacobs@kevinjaybe·
So JD Vance, vice president of the USA, travels to a EU member state days before an election, to complain about election interference by the EU bureaucrats? You can say what you want about the EU but come on man..
English
0
0
1
32
Kevin Jacobs
Kevin Jacobs@kevinjaybe·
Absolutely none. I repeat none of these smaller party models that I have tried come even remotely close to the established players. The quality varies way too much. This is perfectly illustrated in the arena.ai benchmarks.
AlphaSignal AI@AlphaSignalAI

A peanut-sized Chinese model just dethroned Gemini at reading documents. GLM-OCR is a 0.9B parameter vision-language model. It scores 94.62 on OmniDocBench V1.5, ranking #1 overall. For context, it outperforms models 100x its size. 100% open-source. It works in two stages. 1. A layout engine detects every region in a document. 2. Each region gets read in parallel. The model predicts multiple tokens per step instead of one. That's what makes it so fast at small size. It handles things most OCR tools struggle with: > Complex tables and nested layouts > Handwritten text and stamps > Math formulas and code blocks > Mixed image-and-text documents You can run it locally through Ollama. It fits on edge devices with limited compute. Every expensive OCR API just got a free competitor.

English
0
0
0
36
Kevin Jacobs
Kevin Jacobs@kevinjaybe·
Goosebumps watching the tribute to Wawrinka at Monte-Carlo. 🫶🏼🫶🏼
English
0
0
0
41