Arcee AI Trinity Mini and Nano – US based open weight models(www.arcee.ai)

4 pointsby BarakWidawsky2 months ago1 comment

sosodev2 months ago
If the performance is comparable to Qwen3 in practice that's quite impressive.
Half the dataset being synthetic is interesting. I wonder what that actually means. They say that Datology needed 2048 H100s to generate the synthetic data. Does that mean they were generating data using other open weight LLMs? Seems like that would undermine the integrity of a "US based" dataset.
- davidsainez2 months ago
  Why would that undermine its integrity? AFAICT there are a selection of "open" US-based LLMs to choose from: Google's Gemma, Microsoft's Phi, Meta's LLAMA, and OpenAI's GPT-OSS. With Phi licensed under MIT and GPT-OSS under Apache 2.
  - sosodev2 months ago
    Because at that point you don't know where the data came from. You could be training on foreign propaganda without realizing it.
    Presumably they wouldn't be training on synthetic data produced by anything less than a open frontier model and those are almost exclusively Chinese