[hn1986]

Thank you, this looks very good with the 1M context window. Does it use any data (schema, etc) for training?