Originals?
Do you have access to the original versions so they can be converted into a multi-round format with the initial message preserved? I was able to sequence the original conversations as best as possible from this data, but still don’t have the initial user messages for proper context training of most messages. Wondering if the original source was also truncated/sliced or if it can still be fixed.
Do you have access to the original versions so they can be converted into a multi-round format with the initial message preserved? I was able to sequence the original conversations as best as possible from this data, but still don’t have the initial user messages for proper context training of most messages. Wondering if the original source was also truncated/sliced or if it can still be fixed.
The original version is in a weird format, I can update the README.md to point to a google drive containing the original version.
Edit: Updated.