Douglas Hanley

Hmm, interesting. What's the error message when you try accessing one of the GLM routines like `fr.glm`? Also what's the value of `fr.HAS_JAX`? That should be `True` when it's working.

Thanks for the feedback! Good to know that it's working now. Version 1.1 also had some of the non-linear routines, but they way I was importing it then, it would...

Thank you! I don't get to test things much on Windows, so it's good to have some feedback there. I just committed some changes in 936814a that handle things from...

What kind of errors are you seeing with the existing code? Do they occur when you're writing or reading later? In the current codebase, we aren't specifying the encoding, so...

Thanks for the info. It looks like `open` on your system defaults to the `gbk` encoding, which has trouble with unusual unicode characters like `\u2003` (em space). This can be...

Is the SPM preprocessor also replacing accented characters? Seems like we should be able to reuse bits from that. Btw, in case it's useful for folks, I made a little...

That's very helpful @cebtenzzre! Opening a PR with this in a minute.

This seems like a useful idea! As you describe, it seems like a tricky thing to satisfy all the various constraints. On the `llama-cpp-python` side there have also been reports...

It looks like this is actually a tokenization issue. I'm seeing the output of `/tokenize` as being pretty garbled. First, it doesn't appear to be adding an BOS token. We're...

Trying to figure out what's up with `special_tokens_cache` and looking through https://github.com/ggerganov/llama.cpp/pull/3538 for guidance. Most models I'm looking at seem to correctly label special tokens with `token_type`. Do we have...