jamesmf/cclm

Issues

• TypeError: Object of type int64 is not JSON serializable

  #54 opened 3 years ago by jamesmf
  1

• add dgl pretraining task

  #53 opened 3 years ago by jamesmf
  0

• add distillation pretrainer

  #52 opened 3 years ago by jamesmf
  0

• Support lower versions of python

  #46 opened 3 years ago by jamesmf
  1

• dockerfile args for installing deep graph library

  #50 opened 3 years ago by jamesmf
  0

• Add option for Preprocessor to add [CLS] token

  #44 opened 3 years ago by jamesmf
  1

• Make preprocessor aware of downsampling

  #43 opened 3 years ago by jamesmf
  1

• remove CCLMModelBase in favor of Embedder

  #41 opened 3 years ago by jamesmf
  2

• add changelog

  #38 opened 3 years ago by jamesmf
  1

• initial shelf implementation

  #39 opened 3 years ago by jamesmf
  1

• modify MLM pretrainer to accept model architecture args

  #10 opened 3 years ago by jamesmf
  1

• put position embedding back on transformer implementation

  #34 opened 3 years ago by jamesmf
  1

• consider not requiring a preprocessor on a base

  #29 opened 3 years ago by jamesmf
  0

• increase test coverage on pretrainers

  #31 opened 3 years ago by jamesmf
  1

• standardize pretrainer api

  #32 opened 3 years ago by jamesmf
  0

• readme stale

  #21 opened 3 years ago by jamesmf
  0

• logic for picking random substring in MLM pretrainer

  #9 opened 3 years ago by jamesmf
  1

• script for training and persisting pretrained model on wikipedia-en

  #11 opened 3 years ago by jamesmf
  1

• masked character sequence pretrainer

  #20 opened 3 years ago by jamesmf
  2

• start using poetry

  #23 opened 3 years ago by jamesmf
  1

• Add categorical pretrainer

  #17 opened 3 years ago by jamesmf
  0

• Add causal language model pretrainer

  #16 opened 3 years ago by jamesmf
  0

• create evaluation method for MLM pretraining

  #7 opened 3 years ago by jamesmf
  0

• mixed precision support

  #12 opened 4 years ago by jamesmf
  0

• Implement freezing for pretrainer and test it for base

  #6 opened 4 years ago by jamesmf
  0

• new tokenizer

  #2 opened 4 years ago by jamesmf
  5

• dependencies in setup.py

  #3 opened 4 years ago by jamesmf
  0

• decide on a default dataset

  #1 opened 4 years ago by jamesmf
  0