@rustlang @google Past: #rstats, @bazelbuild, @grpcio and @netty_project. Opinions my own.
Munich, Bavaria
Joined January 2013
- Tweets2.2K
- Following828
- Followers424
- Likes7.7K
Nice!!!
I just opened a pull request to add support for Cloudflare KV caching and Cloudflare D1 memory storage in @langchain JS 🤖
This allows you to save context and past history - example code below!
github.com/langchain-ai/lang…
jakob buchgraber retweeted
I just opened a pull request to add support for Cloudflare KV caching and Cloudflare D1 memory storage in @langchain JS 🤖
This allows you to save context and past history - example code below!
github.com/langchain-ai/lang…
jakob buchgraber retweeted
pay attention to this trend where governments will push for more projects to adopt modern memory safety focused languages – at Austin, TX
Breaking News: The National Cybersecurity Strategy Implementation Plan was just published by the @WhiteHouse. Excited to see investing in memory safety as a strategic objective.
buff.ly/3pRCw63
jakob buchgraber retweeted
Probably the best thing you'll see today.
In 2017, a group of developers hilariously competed for who could create worst volume control interface in the world.
The results 🧵
1/22
jakob buchgraber retweeted
This is very cool: it has become a lot easier to read academic research.
If you open a paper in Microsoft Edge (I know, I know), the Bing AI sidebar can read the PDF and you can ask it questions about the paper. (I checked & the results seem high-quality, but be careful with AI)
jakob buchgraber retweeted
Really striking how every AI hype tweet is explicitly trying to induce FOMO. "You're getting left behind. All your competitors are using this. Everyone else is making more money than you. Everyone else is more productive. If you're not using the latest XYZ you're missing out."
jakob buchgraber retweeted
If you ask everybody in an organization what percentage of the organization's success they're personally responsible for, you'll wind up with a total of about 300 percent. That's just the reality, and it shows why you must be precise in attributing specific results to specific people's actions. Otherwise, you'll never know who is responsible for what – and even worse, you may make the mistake of believing people who wrongly claim to be behind great accomplishments. #principleoftheday
In a coding world we used to express costs in relation to engineers compensation. Going forward we will measure compensation in GPU hours👌👍. If your model costs billions to train… 🤣.
jakob buchgraber retweeted
When you speak German, this video by Harald Lesch about eFuels is fun to watch. Harald can't quite believe that something so inefficient as eFuels is presented as a solution for car mobility.
ht @WulfHemmerle
youtube.com/watch?v=MnrudYCz…
jakob buchgraber retweeted
"Please, tell me again why you need a car..."
🛴
xitter.cf/i/status/1645370976157…
jakob buchgraber retweeted
I agree with Macron. I want Europe to become the world's "third world power." Another democratic great power in the world would benefit U.S. security, economic, and moral interests. But to become a great power, one has to act like one, and spend like one, esp on defense.
jakob buchgraber retweeted
The *behavior* (but not the capability) of a model depends largely on its final fine-tuning objective. GPT 3.5 & 4 using instruction tuning and RLHF, which do, in fact, reward "talking like a human". Increasingly models use model-designed criteria, making them even more GAN-like.
Couldn’t agree more! Dr. Mike Pound is numero uno.
Various public figures and scientists have recently released a letter calling for a pause in research on large language models. Confused about all this? Let's take a look.. 1/
futureoflife.org/open-letter…
jakob buchgraber retweeted
This explains very nicely & clearly how large language models are formulated from a probabilistic perspective.
Google Robotics papers:
"Grounded Decoding: Guiding Text Generation with Grounded Models for Robot Control"
jakob buchgraber retweeted
i summarized and compiled all of the literature i feel is relevant for catching up on the state of ai in the lm-flavoured space. everything links to directly to the pdf (not the arxiv home)~ it covers 22 models along with two dozen other techniques
kipp.ly/blog/transformer-tax…
