ThomasHoppe

@ankurankan You wrote "Since the BayesianModel class can represent a NoisyOr model structure as well" I ask myself how this is done? I usually use BayesianNetworks, but now I need...

Under Python 3.12.4 and rdflib 7.0.0 vanished

> Same error > > Solved by adding --use_cached_embd option. > > This is the codes in train.py, I think it should do something generate embds when this option is...

> [#33](https://github.com/microsoft/KBLaM/pull/33) - I think this PR solves the dataset-loading? Please let me know if that's not the case. Yeah, this information was missing in the ReadMe. Unfortunately the information...

I found the same last week, but hadn't the time for posting this isssue. It is a bad idea to hard-code the step size. I think, it would be better...

> For the notion of step, it means one gradient descent step, notice that here we intentionally adopt the notion of "number of step" rather than the number of epochs...

> How many steps are required to achieve KBLaM's findings? Am I correct in understanding the whitepaper used 20,000 steps? > > I'm trying to run eval on 400 steps...

Answer to 1. The total_steps needs to be a multiple of the save_intervall in the train.py code.

> I believe this is solved by sticking with huggingface 4.46.0 in [#40](https://github.com/microsoft/KBLaM/pull/40) . Could you let me know if that's fixed things? Well, I am not sure. I think...

Ok downgrading to transformers=4.46.0 seems to solve the problem above. However, afterwards I get another error: `---> 23 from kblam.utils.data_utils import aug_row, generate_multi_entity_qa 24 from kblam.utils.eval_utils import ( 25 instruction_prompts,...