My undergraduate thesis was completed almost a decade ago and is not available online, nor was any publication made from the findings. I would like to share it online, but looking back on it I notice many grammatical errors and conclusions that make less sense to me after gaining research experience. Is it unethical to make considerable changes to the text, or should I produce the document as it is to provide an honest representation of my early work?
IMHO it depends on why you want to make it available.
If your primary reason is that you want to make information available more publicly, you could produce a "revised edition" of the thesis and add a preface stating that this is a substantially revised and edited version of your thesis - just like other books do. If you just correct typos and grammar, just say that.
There is also the possibility to comment on the conclusions you don't see like that any more in a way (formatting, footnote) that makes clear that this is a point where your opinion now differs. (like a commented edition of some work).
If your primary reason is to give the text your online CV refers to, then leave it unchanged.
Is it unethical to make considerable changes to the text
No, it's not unethical, but if you make changes it would be useful to include a note or a preface disclosing that that is an amended version of your original thesis, and if your conclusions have changed considerably after further experience, you can explain why.
You could publish both versions, so readers can even
diff them if they wanted to.
You'd probably want to direct readers to the version you think is better / easier to read (i.e. the revised one). You should still write a short summary of why you updated it and what the changes were, but that section can link to the original.
It's probably fine to correct grammatical mistakes in the original, but any changes to how the ideas are presented should only go into the revised version.
If you want this to be an accurate representation of your undergraduate work, then you should leave the suspect conclusions intact. There's something to be said for the logic of inexperience.
If I were doing this, I would edit the text for grammar (including spelling, typos, etc.) and possibly form (section headings, page layout, etc). I would also add a preface discussing the changes, pointing out that they are for readability and do not affect the content of the work. I would also comment on the content in the preface, discussing how I would amend the old conclusions based on new experience.
It might also be of interest to address why this is a worthwhile endeavor at all in the preface. Was there some new finding that sheds light on this old work? Is the old work a step in the direction of a recent technique?