Ehsan M. Kermani

I could successfully build using the **nightly libcore: rustc 1.27.0-nightly (0b72d48f8 2018-04-10)** with some changes. My related system specs: `Ubuntu 16.04, LLVM 6.0, gcc 5.4, CUDA 8` 1) _Remove_ the...

@TGM I just updated my prev post. Probably you should modify the nvptx64-nvidia-cuda.json for windows! also get rid of Xargo.toml and `https://github.com/japaric/` as I said before.

What environment are you installing it in? I have got the similar issue in docker `python:3.10-slim-buster` base image and adding `apt-get install --no-install-recommends -y build-essential gcc` fixed it, however locally...

@Musa-Sina-Ertugrul are you still working on it? The previous PR was closed.

* To use `chain = load_qa_with_sources_chain(...)`, first you need to have an index/docsearch and for `query` get the `docs = docsearch.similarity_search(query)` to use `chain({"input_documents": docs, "question": query}`. * `RetrievalQAWithSourcesChain` is...

It's the issue with [uvloop doesn't support windows yet](https://github.com/MagicStack/uvloop/issues/14) and not langchain. Other option is to use docker.

This is how it is [implemented](https://github.com/facebookresearch/faiss/blob/d87888b13e7eb339bb9c45825e9d20def6665171/faiss/Index.h#L274) in faiss i.e. to move to index1 for memory efficiency. If you need index2, you can `copy.deepcopy(index2)`.

The `model_path` should refer to a saved torchscript roberta model. Is that so?

Thanks for reporting! We are looking into this.

Thanks for reporting! this seems to be an isolated instance. Could you try again and let us know if it persists?