Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Incorrect architecture specified for ANCE DPR context encoder model

See original GitHub issue

In the config for your ANCE DPR context encoder in HuggingFace model hub, the architecture is listed as DPRQuestionEncoder but it should be DPRContextEncoder. This causes a failure when using “reflection” to check the correct model class type to use.

See: https://huggingface.co/castorini/ance-dpr-context-multi/blob/main/config.json Related: https://github.com/huggingface/transformers/issues/13670

Issue Analytics

State:
Created 2 years ago
Comments:7 (5 by maintainers)

Top GitHub Comments

1reaction

MXueguangcommented, Sep 21, 2021

It shouldn’t affect results, but I’ll rerun to make sure it works.

1reaction

MXueguangcommented, Sep 21, 2021

Yes, @lintool @joshdevins, thank you for catching the issue, I’ll update the model on hugging face.

Top Results From Across the Web

DPR AutoModel loading incorrect architecture for ... - GitHub

Loading a DPR context encoder DPRContextEncoder using AutoModel.from_pretrained is actually loading DPRQuestionEncoder instead, and later fails. Steps to ...

DPR - Hugging Face

It is used to instantiate the components of the DPR model according to the specified arguments, defining the model component architectures.

PyTerrier Documentation

PyTerrier uses some of the functionality of the Java-based Terrier IR platform for indexing and retrieval functionality.

arXiv:2212.09114v1 [cs.CL] 18 Dec 2022

In this paper, we propose a curriculum learning approach to bridge the gap between training and inference for passage retrieval using dual-cross ...

Passage Retrieval | Papers With Code

Open-domain question answering relies on efficient passage retrieval to select candidate contexts, where traditional sparse vector space models, such as TF-IDF ...