Keshav Santhanam
Keshav Santhanam
Hi @ebritoc, it seems the main training loop is not being executed for some reason - it's possible that there is some data missing that is preventing the data loader...
That appears to be the correct format for the triples in terms of having a list of lists, each with three integers - but do the qids and pids correspond...
Sorry for the delay in getting back to you, but I tried reproducing this on my end and I think the issue with the minimal example at least is that...
Ah could you try organizing your `triples.test.jsonl` as follows? ``` [0,0,1] [1,1,0] ```
Hi @ashleemilton, I haven't tried working with CUDA through WSL but it seems like there are some setup steps that NVIDIA lists here in case you haven't seen these before:...
Ah it seems from this thread that single-gpu support on WSL is only available on NCCL version 2.10.3, and multi-gpu support is only available with NCCL version 2.11.4: https://github.com/NVIDIA/nccl/issues/442 I...
Closing due to inactivity, please re-open if this is still an issue.
Hi @lalenzos, this is feature we are planning to add soon - stay tuned!
Agree with @okhat, this is very strange - I'll look into it.
This is correct, FAISS is only used for the k-means computation.