unicamp-dl/mMARCO

Issues

• Larger 13B model underperforms BASE model, any idea why ?

  #22 opened 10 months ago by cramraj8
  3

• What are the hyper-parameters values used to train mT5-base models?

  #21 opened 10 months ago by cramraj8
  3

• mt5-base model (unicamp-dl/mt5-base-en-msmarco) throws 'nan' outputs

  #20 opened 10 months ago by cramraj8
  2

• How to get cross-lingual triplet pairs ?

  #16 opened 10 months ago by cramraj8
  1

• What is the license of the dataset?

  #19 opened a year ago by PhilipMay
  1

• Language for Pyserini BM25 indexing and retrieval should be a two-letter code ('pt' instead of 'portuguese')

  #18 opened a year ago by marcospiau
  2

• problem in japanese data

  #17 opened a year ago by guijuzhejiang
  1

• Google Cloud bucket not working

  #8 opened 3 years ago by mirfan899
  17

• How to download all the datasets?

  #15 opened 3 years ago by RiverDong
  0

• Loading mMARCO using HuggingFace Datasets API

  #13 opened 3 years ago by atamborrino
  2

• Translation issues in dev query file?

  #7 opened 3 years ago by jobergum
  3

• How to finetune the model on mMarco

  #9 opened 3 years ago by shunyuzh
  3

• How to reproduce the BM25 scores ?

  #2 opened 3 years ago by LoicDagnas
  2

• Question on reranking with multilingual-MiniLM

  #5 opened 4 years ago by jobergum
  2

• Why .txt files (queries.dev, queries.train) present in Portuguese whereas .tsv for rest?

  #3 opened 4 years ago by thakur-nandan
  2

• Portuguese missing (& other languages)

  #1 opened 4 years ago by seanmacavaney
  2