Aleph Alpha

@Aleph__Alpha

We create specialized large language models for a sovereign Europe. Join us: jobs.ashbyhq.com/AlephAlpha #artificialintelligence, #writtenbyahuman

Heidelberg, Germany

Aleph Alpha
@Aleph__Alpha
7h
Hot off the press: We are becoming the first transatlantic sovereign AI solution together with our partner @cohere. More talent, more compute, and more innovation power to offer trustworthy AI at the security level that governments and enterprises need – across the globe.
2
Aleph Alpha
@Aleph__Alpha
Sep 10
Training and evaluating LLMs is hard: it’s a multi-stage process, with different methods, data, and evals at each stage. In our new post, @SohirMaskey and @_wsascha_ study when intermediate evals become predictive of final model quality in a 30B MoE. 🧵tinyurl.com/8zh8hvjp
1
Aleph Alpha
@Aleph__Alpha
Sep 8
Sauerkraut, not burgers - why German LLMs need German data. We wanted to train a model that truly speaks German: grammar, cultural context, customer use cases. Turns out the training data you need for that mostly... doesn't exist yet. A thread on how we fixed that 🧵
2
Aleph Alpha
@Aleph__Alpha
Aug 19
Your model answers 9 of 10 questions correctly about a contract you gave it to check, but you don’t know which one is wrong, so you check all 10. LLM training rarely rewards "I can’t answer": abstaining pays less than guessing, so models guess plausibly, & that’s hallucination.🧵
2
Aleph Alpha
@Aleph__Alpha
Aug 13
Ever wondered why LLMs read fast but generate slowly? Inside GPU inference: tinyurl.com/mr3mshr8
1