3 ms·
I struggled with the same problem. What's interesting with LLMs, is that they will augment the data in the paper with their general knowledge. So you can talk t
by addandsubtract 3y ago
I struggled with the same problem. What's interesting with LLMs, is that they will augment the data in the paper with their general knowledge. So you can talk to them like you would with a colleague or mentor (but without the shame of asking dumb questions).
- 3abiton 3y agoBut are there currently any good local models that can provide good conversation/support of a recent paper? I feel the level of hallucinations will be huge. Even with gpt4 with pdf2text, it tends to hallucinate a lot. And Gpt4 is the best LLM model so far.