Kurt Shuster
Kurt Shuster
We typically pass the topic in as the very first message in the dialogue history; so, your first turn would instead be: ```python first_turn = "\n".join( [ "Movies", "your persona:...
Each "your persona: " line is a persona you would want the bot to emulate. If you wanted to specify the partner's persona, you would have `"partner's persona: "`
yes, you can do exactly as you've described. it remains to be seen how effective this would be, depending on the model used
any conversational model would be a reasonable choice, bb2_400m is good
fine-tuning on new data while preserving knowledge of older capabilities is an open problem in language modeling. One approach would be to incorporate some original training data within the fine-tuned...
Yes, we offer several ways of doing this, via the `--rag-retriever-type`, such as using neural retrieval over document embeddings, of TFIDF over document sets. [Please see the relevant README for...
Only to the extent the bot can fit the tokenized input into it's context; that is determined by the `--truncate` / `--text-truncate` flags if you're curious about bb2, that is...
The real factor is the number of position embeddings the pre-trained models used; the 3B model just so happened to use 128 positions while the 400M used 1024 (3B is...
Ahh, when I said there were no length restrictions I meant that ParlAI itself could handle any length; the models are bound by their truncation length