103 karma · joined February 10, 2024
There is no distributed execution of a single LLM across multiple GPUs using tensor parallelism here. Instead, it relies on task-level parallelism-breaking down a large workload into independent subtasks that run across separate machines. For example, generating embeddings for a million documents.
But: Hyper, Bytes, Mio, Axum, Prost
Overall, after working with Warp, Ghostty+Herdr, and tty7, tty7 has felt the most comfortable so far.
tty7 has its drawbacks too - the biggest one for me is the lack of support for switching tabs with `Cmd+Shift+[` and `Cmd+Shift+]`.
Firefox is steadily losing market share, and any attempts to do something about it are met with negativity. The 2-4% of users who use it care about their privacy. But they are not being deprived of it; the AI tab is optional, and no one is removing the regular tab. (Of course, it would be better if they allowed the integration of local models or aggregators, such as Openrouter, Huggingface...)
Meanwhile, developers continue to ignore Firefox, testing only Chromium browsers. Large companies are also choosing the Chromium engine for their browsers.
Perhaps if they implement this functionality conveniently, more average users will use Firefox.