Create your own GitHub profile
Sign up for your own profile on GitHub, the best place to host code, manage projects, and build software alongside 50 million developers.
Sign upHighlights
- Arctic Code Vault Contributor
Pinned
631 contributions in the last year
Activity overview
Contribution activity
August 2020
tholor has no activity
yet for this period.
July 2020
Created a pull request in deepset-ai/haystack that received 1 comment
Upgrade to new FARM / Transformers / PyTorch versions
Upgrading to latest FARM (0.4.6) which comes with PyTorch 1.5.1 and Transformers 3.0.2 This means: Making FARMReader more powerful & stable (e.g. …
+2,083
−61
•
1
comment
- Add num_processes to reader.train() to configure multiprocessing
- Fix return type of EmbeddingRetriever to numpy array
- Add eval for Dense Passage Retriever & Refactor handling of labels/feedback
- Fix format of /export-doc-qa-feedback to comply with SQuAD
- Refactor DPR for latest transformers version & change init arg `gpu` -> `use_gpu` for DPR and EmbeddingRetriever
- Fix multi gpu training via Dataparallel
- Harmonize meta data handling across doc stores
- Add more tests
- Add dummy retriever for benchmarking / reader-only settings
- Update tutorials
- Tutorial for Dense Passage Retriever
- WIP: Add DPR Training
- Fix type of query_emb in DPR.retrieve()
- Add method to update meta fields for documents in Elasticsearch
- Remove meta field when writing documents in Elasticsearch
- Fix Evaluation Dataset
- Fix id for documents returned by the TfidfRetriever
- Remove mutation of documents in write_documents()
- Remove mutation of results dict in print_answers()
- Added Support for Docx Files
- Move document_name attribute to meta
- Fix embedding method in FAQ-QA Tutorial
- Fix Dockerfile to build successfully without models directory
- Update docstring for embedding_field and embedding_dim
- Fix naming of offset in answers of TransformersReader (for consistency with FARMReader)
- gitignore fix
- Docker available for TransformsReader Class
- Add response time in logs
- Add success response for successful file upload API
- Adjust PDF conversion subprocess for Python v3.6
Created an issue in huggingface/nlp that received 3 comments
Is there a way to download only NQ dev?
Maybe I missed that in the docs, but is there a way to only download the dev set of natural questions (~1 GB)? As we want to benchmark QA models on…
3
comments