Josh Devins
Josh Devins
Might want to consider using hiredis-rb instead of requiring hiredis to be installed on the OS before hand. Can look into this if I have time too, but for now,...
Turned native: true, and Recommendify would fail silently w/ Jaccard similarity instead of throwing some error when native extensions are failing. (if not built, I do get a proper error...
Today we just use `transformers.AutoModel.from_pretrained` to load Hugging Face models from the hub, but this will only work for public models. We should expose flags and support for accessing private...
With the recently added Dockerfile (#405) for building a Docker image from the current/local eland, we introduce the ability to run eland installed scripts without installing eland or any prerequisites....
First, a thank you. The paper and repo have been fantastic resources to help conversations around out-of-domain retrieval! Second, a feature request. I think it would be very interesting to...
Just wondering if you could add somewhere a description of your MS MARCO submission "C-COIL + RoBERTa" from 14/07/2021. Which modifications to COIL have you made and what was the...
Line 158 of text2people: 'princess' => 'm' should be 'f' (I would say)
When we install `sentence-transformers` in our `pytorch` [target](https://github.com/elastic/eland/blob/main/setup.py#L62), we pull in all of the transitive dependencies of that library which [include many libraries](https://github.com/UKPLab/sentence-transformers/blob/master/setup.py#L21-L32) that we will never need. Excluding these...
Today we rely mostly on unit testing for the PyTorch/NLP model import testing. We perform large scale testing as part of other components like Elasticsearch, but we often find bugs...
Dynamic quantization of PyTorch models has proven to be a challenge for two reasons. (1) Dynamic quantization ties the traced TorchScript model to a particular architecture and makes it non-portable....