xlang-ai/UnifiedSKG

Issues

• (question on paper) knowledge source used for WQSP?

  #45 opened 9 months ago by Solomon-Han-00
  0

• Help wanted! confused about the prefix_spider_with_cell_value.cfg in the T5-base configuration and T5-Large configuration.

  #44 opened 10 months ago by kanseaveg
  2

• Help wanted! What is the purpose of adapter tuning in the code? Why does the adapter need to introduce a certain number of virtual tokens?

  #43 opened 10 months ago by kanseaveg
  3

• Error on colab

  #42 opened a year ago by Tizzzzy
  2

• NQ-Tables test set evaluation

  #39 opened 2 years ago by yhifny
  2

• Can not reproduce the results of spider with T5-base

  #41 opened a year ago by xufangzhi
  3

• NCCL version

  #12 opened 3 years ago by cdhx
  12

• CUDA OOM with DeepSpeed stage 3 for 11B T5 model

  #38 opened 2 years ago by SOUMAJYOTI
  1

• Knowledge Graph as Input: question-specific subgraphs

  #9 opened 3 years ago by cmavro
  7

• Model checkpoints and outputs for T5-3b

  #37 opened 2 years ago by carlos-gemmell
  1

• Result verification - could you please give a pointer to the model used to obtain SPIDER results in the paper?

  #36 opened 2 years ago by migakol
  1

• Help with reproducing T5-3b number on Spider

  #35 opened 2 years ago by yeliu918
  2

• Is there a unified entrance to reproduce all paper experiments?

  #14 opened 3 years ago by ShaneTian
  2

• the original prefix-tuning corresponds to the models in the directory `models/prompt`?

  #34 opened 2 years ago by Chacha-Chen
  8

• About training time

  #33 opened 2 years ago by LucyFann
  2

• Regarding tuning the prefixes

  #32 opened 3 years ago by base-y
  5

• How to use fp16

  #31 opened 3 years ago by JBoRu
  7

• Getting the multi-task prefix weights

  #30 opened 3 years ago by base-y
  4

• Question on Prefix tuning code

  #28 opened 3 years ago by base-y
  6

• How can I specify PLM folder

  #19 opened 3 years ago by puraminy
  8

• How can I train the multi-task models?

  #29 opened 3 years ago by jifan-chen
  6

• The different results between eval mode and test mode.

  #26 opened 3 years ago by eyuansu62
  14

• prefix tuning with t5-3b

  #27 opened 3 years ago by zluw1117
  11

• DDP training default

  #25 opened 3 years ago by eyuansu62
  8

• [Deprecated] Size mismatch when setting knowledge_usage to separate

  #24 opened 3 years ago by puraminy
  6

• WikiSQL model producing only one token as output

  #20 opened 3 years ago by MatriXcel
  2

• [Deprecated] Separate setting: OverflowError: cannot fit 'int' into an index-sized integer

  #23 opened 3 years ago by puraminy
  7

• The saved model folder do not contain config.json.

  #22 opened 3 years ago by eyuansu62
  5

• [Bug] When running without WandB, the `wandb_run_dir` is None

  #21 opened 3 years ago by ShaneTian
  2

• Can I use the project for a text to text task?

  #18 opened 3 years ago by puraminy
  1

• TypeError: get_hash() missing 1 required positional argument: 'use_fast_tokenizer'

  #17 opened 3 years ago by puraminy
  5

• TypeError: add_code_sample_docstrings() got an unexpected keyword argument 'tokenizer_class

  #16 opened 3 years ago by puraminy
  1

• About prefix tuning

  #15 opened 3 years ago by drxmy
  2

• Checkpoints for models trained on Spider

  #13 opened 3 years ago by awasthiabhijeet
  3

• Questions about MultiWOZ and SMD (KVRET)

  #2 opened 3 years ago by ShaneTian
  7

• GPU and batch size setting

  #11 opened 3 years ago by cdhx
  11

• huggingface model download failed

  #10 opened 3 years ago by cdhx
  5

• Hyperparameters for all the datasets

  #8 opened 3 years ago by gao-xiao-bai
  2

• BLEC metric for SQL-to-Text

  #6 opened 3 years ago by awasthiabhijeet
  3