AI something @Aleph__Alpha. Post training LLMs
Heidelberg, Germany
Joined August 2018
- Tweets889
- Following2K
- Followers263
- Likes3.6K
Letting sandbox creation choke GPU utilization is one of the costliest mistakes in async RL
We learned this the hard way during Kolibri's development. Tackling this and several other scaling challenges helped us reach 25k concurrent sandboxes & millions of sandboxes overall
Blog link: aleph-alpha.com/en/blog/why-…
The move from external services to self-hosting didn't stop at sandboxes. We followed a similar path with other environment infrastructure, adding more capabilities along the way.
You can see how the stack evolved
We started with external sandbox providers but quickly moved to self-hosting our own sandbox service
This gave us much more control over the infrastructure and better observability into what was happening under the hood
Murali Kondragunta retweeted
Every architectural choice has a cost. Choose where to pay it.
@alessio__serra and @PitNeitemeier derive from first principles how design choices allocate parameters, FLOPs and KV cache, compare open models, and share some of the tooling we used to design Kolibri.
During its development, Kolibri surprised us in ways we didn't expect with reward hacking & escaping sandboxes. We were a little proud but also increased our sandbox isolation
For an 80B-A3B model, it is unexpected. All the details are in "Lessons learned" section in tech report
This is so amazing to see! Love open source for this!
Btw, we didn't train Kolibri on blender or artifact generation. So good to see that model generalized to do this
Looking forward to more such experiments :) :
huggingface.co/Aleph-Alpha/K…
💥 Kolibri is HERE, and it's Europe's answer to open-weight AI
🐦 Aleph Alpha's new German/English model: 78B total parameters, only ~3.5B active per token
🔹Up to 1M token context window
🔹Apache 2.0 license, run it on 2x H100 or 1x H200
🔹Built on a "training as code" pipeline: the big run is just the last step after hundreds of small experiments
🔹Strong on math and German-language tasks (benchmarks are the company's own claims)
🔹Tested hands-on: a full Blender + Godot trampoline park game, a cinematic doner kebab in pure HTML, and a brutal Hegel German-English philosophy challenge
🔥Watch the full video below:👇
youtu.be/eED06GqChT0?si=5MB_…
@Aleph__Alpha
Murali Kondragunta retweeted
People saying "it's worse than Qwen3.6 35B" don't understand that support comes from the bottom. If we don't support the local AI ecosystem, no matter the stage, no matter the location, we're always going to fall prey to "yeah but America/China does it better". Those who tried to build something the day GPT tool calling came out (🖐) know very well the pain of getting malformed tool calls and the model going haywire with the most trivial requests. That was just 3 years ago. Today you can run an even better model on a laptop.
I'm a huge fan of frontier labs, they are the pioneers of what we have today and I'll forever be grateful, but local AI is one of the most important (and frankly exciting) challenges of this decade. We're all in it together. Supporting each other is the least we can do. So congrats on the release! 🇪🇺
Thanks Max! ❤️
Max played an instrumental role in building the core training teams before leaving to found Construct Lab
Try out Kolibri at
kolibri.constructlabs.com
Big congrats to my friends at Aleph. Glad I was still around when this team formed and got to see how much work went into building it end to end.
kolibri.constructlabs.com
We will offer inference for free for the next ~48 hours.
Hit me up if you are interested in bringing this into production or post-training it.
Murali Kondragunta retweeted
Big congrats to my friends at Aleph. Glad I was still around when this team formed and got to see how much work went into building it end to end.
kolibri.constructlabs.com
We will offer inference for free for the next ~48 hours.
Hit me up if you are interested in bringing this into production or post-training it.
Our model, Kolibri, is out! 🐦
Incredibly proud of how much we've scaled in just a few months
RL is as much an infrastructure challenge as a modeling one. We've scaled our RL environments and their infra to support 20K+ concurrent sandboxes per training run and 100K+ sandboxes overall. More details in the report
A massive engineering effort behind the scenes. Proud of what we've built!
Murali Kondragunta retweeted
I see people being sarcastic but there is a real demand for it: if you want to embed models in industrial processes you need to ease audit and doing local language (and even more local professional register) remove frictions.
Murali Kondragunta retweeted
Reasoning models think in English, even on German prompts. We asked what it costs to make one think in German. Answer: a valley. Small doses of German reasoning data hurt, large doses mostly recover.
