facebookresearch/DPR

Issues

• Missing required positional arguments

  #237 opened 2 years ago by grossmanm
  6

• 'dpr_all_documents' is not defined

  #249 opened a year ago by golubovic
  2

• nanoDPR: simple replication of DPR

  #251 opened a year ago by Hannibal046
  0

• mismatch between encoded results and wiki passages

  #250 opened a year ago by Hannibal046
  0

• A possible bug: Wrong hard_neg_ctx_indices

  #248 opened a year ago by zhiyuanpeng
  1

• Training DPR with different language

  #240 opened 2 years ago by shahad2099
  1

• an error occurred in Retriever inference

  #211 opened 3 years ago by juyoung228
  2

• How to enter model_file in dense_retriever.py?

  #241 opened 2 years ago by tkabir1
  1

• Having error while running train_extractive_reader.py

  #242 opened 2 years ago by tkabir1
  1

• IndexError: Dimension out of range

  #213 opened 3 years ago by mary-octavia
  12

• error about python dense_retriever.py

  #245 opened 2 years ago by CodingPeasantzgl
  0

• IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)

  #225 opened 2 years ago by ZiluLii
  3

• question: reader loss

  #244 opened 2 years ago by PaulLerner
  0

• Retriever does not train properly on a custom dataset with empty negative_ctxs and hard_negative_ctxs

  #243 opened 2 years ago by yu-rovikov
  2

• Data from retriever is not working with reader

  #229 opened 2 years ago by UmerTariq1
  1

• Can we get wiki page url by passage id?

  #239 opened 2 years ago by heya5
  0

• About dataset generation

  #238 opened 2 years ago by Heisenberg-Yin
  0

• Unable to reproduce table 3 (Gold-127N) from scratch

  #236 opened 2 years ago by xhluca
  1

• Passages with multiple answer spans

  #233 opened 2 years ago by TZeng20
  0

• `cosine_scores` defined in biencoder.py does not work

  #232 opened 2 years ago by xhluca
  0

• Question in training reader model.

  #231 opened 2 years ago by gaishun
  1

• Doubt regarding all_gather_list in case of DDP

  #230 opened 2 years ago by bhattg
  0

• RuntimeError: Missing key(s) in state_dict: "encoder.embeddings.position_ids"

  #228 opened 2 years ago by NehaNishikant
  0

• [Feature Request] Allowing unix-style pattern matching in `download_data.py`

  #227 opened 2 years ago by xhluca
  0

• How do the passage embeddings use the 'title' of the passage

  #224 opened 2 years ago by jigsaw2212
  1

• Unable to install dependencies

  #223 opened 2 years ago by jigsaw2212
  2

• how to test the prediction?

  #217 opened 3 years ago by liuchiente
  0

• Question about Wikipedia Corpus Preprocessing

  #222 opened 2 years ago by manveertamber
  0

• NQ Dataset

  #221 opened 3 years ago by varshakishore
  1

• Retriever Training Error with train_dense_encoder.py

  #220 opened 3 years ago by AIML-YOGI
  1

• Seek Wikipedia corresponding passage ID or dump

  #219 opened 3 years ago by jzhoubu
  1

• Loss across the nodes....[multi-gpu training]

  #218 opened 3 years ago by gabinguo
  2

• Hard Negative Indices Issue

  #212 opened 3 years ago by khuangaf
  3

• How to get ctx_encoder from the checkpoint?

  #216 opened 3 years ago by ali-abz
  2

• error when python dense_retriever.py

  #214 opened 3 years ago by ztx313
  1

• Why does the tokenization part need to set the last token position to sep_token_id?

  #210 opened 3 years ago by kwang2049
  1

• Cannot fully reproduce the embeddings with the nq-single checkpoint

  #208 opened 3 years ago by kwang2049
  1

• About 'speech_data.py' file in dpr/data

  #209 opened 3 years ago by xyltt
  1

• Converting checkpoints to HuggingFace format

  #207 opened 3 years ago by tanay2001
  1

• Importerror

  #206 opened 3 years ago by Facico
  2

• Question about reader training

  #204 opened 3 years ago by TZeng20
  1

• Weird discontinuities in training

  #205 opened 3 years ago by emilysilcock
  2

• TypeError: string indices must be integers

  #198 opened 3 years ago by Suki-m730026038
  1

• Question: Is there a way to reduce GPU's ram consumption?

  #202 opened 3 years ago by ali-abz
  2

• Question on 'gradient_accumulation_steps'

  #201 opened 3 years ago by jzhoubu
  2

• Question about answer quality quantification

  #200 opened 3 years ago by churinga
  4

• Questions about the path of the dev files

  #199 opened 3 years ago by unw9527
  2

• about fix_ctx_encoder

  #196 opened 3 years ago by wuyaoxuehun
  1

• About the NQ model trained with mined hard negatives.

  #193 opened 3 years ago by shunyuzh
  1

• jsonlines is missing as declared dependency in setup.py

  #194 opened 3 years ago by jobergum
  2