Models performed well without, and a little better with, the images. In one case, our no-image model outperformed ALL of the current models on the chest x-ray benchmark—including the private dataset—ranking at the top of the leaderboard. Without looking at a single image.
🤯🤯
Join us on June 6 at 6pm in SF for an evening dedicated to ETL and data processing tools and their impact on data reporting.
@AirbyteHQ and I will moderate a panel featuring data veterans such as Michel Tricot (CEO at Airbyte), @floydophone (CEO at Dagster Labs), @DavidKrakov (CEO at Honeydew), @peterfishman (CEO at Mozart Data), and data celebrities like @_abhisivasailam and @SeattleDataGuy.
We’ll discuss ETL tooling, data quality KPIs, the future of ETL, and root cause analysis using metrics decision trees.
With such great speakers rarely in town, this is an event you won’t want to miss!
Link to signup below.
In my view, semantic models are the means to turn Data into Information. LLMs do not do well with data that’s in rows and columns, but maybe LLMs can be taught to work with Semantic models because they won’t have to deal with how different tables or data points are related to each other, but instead focus on Measures, and one could argue that in some sense Measures are the ultimate expression of Information for tabular data.
Data and analytics are going through a massive wave of tooling that is rapidly accelerating:
▪ What is the right way to set up your reporting today?
▪ Should you use semantic layers?
▪ Do you need dbt?
▪ Which data setup best supports a hundred metric dimensions?
▪ Should you prioritize reporting flexibility or consistency?
I invited @DavidKrakov, the co-founder of YC-backed Honeydew (YC W23), to share his strategies for automating metrics and highlight scenarios where each approach proves most effective.
Read about the pros and cons of every reporting setup - https://t.co/40MIxFIOfb
One of best essays i've ever read on how & why data works.
Weekend plans - digesting rhe whole essay collection (thanks Cedric for giving it as an epub)
The way most people think about data is simply too ... different; too naive perhaps, to reconcile it with the ideas of the WBR.
So I decided to write about those ideas directly FIRST, to set up for a sequence about the WBR.
This essay is out now: https://t.co/ljDwpFDI59
And whatever the method, one piece of advice: go deep.
For a B2B SaaS Growth Model to really be valuable, you have to go far deeper than growth accounting identities, and into the sub-Equations that drive those identities.
Most cos need far more metrics than they think!
Let me get this straight:
1. They spotted a father and a child.
2. Decided to pull out a camera and hit record.
3. Shot the father dead in front of the child's eyes.
4. Dragged the horrified poor child into an underground hellhole.
5. Showed to the child the video of his own father getting murdered again and again.
6. FOR 2 FULL MONTH !!!
7. Forced the child at gunpoint to wave and smile when released.
8. The child go back to an empty home.
9. And have to live with this for the rest of his life.
10. Some people online somehow are rooting for the terrorists.
Right.
#HamasislSIS. Full stop.
THREAD: Another core Gaza lie is that Israel controlled everything that went into Gaza, like caloric intake & Gazans thus oppressed. It’s all lies. Border with Egypt was not controlled by Israel & Gaza brought in ALL items with 3,300 truckloads EACH month. Read for the facts: 1/
60 Minutes has been airing on CBS for more than 55 years. I’m not sure that the award-winning program has ever told a story as dramatic as the rescue of the Tibon family of Nahal Oz. @amirtibon
שרשור עם הפניה בסופו.
זו סבתא חיה. היא לא סבתא שלי. היא סבתא של תומר, אלון וענבר (ועוד מלא נכדים), שהם הילדים של אסתי ואלעד, חברים מאוד מאוד קרובים שלנו.
סבתא חיה גרה בקיבוץ ניר עם. היא בת 77 והיא אישה מצחיקה שאוהבת לחיות את החיים בדרך שלה. לפני שיצאה לפנסיה היא היתה מורה לתנך >