adapter-hub/adapters issues and pull requests

#100 - All the Colabs with adapter training fail due to tensors being on different devices

Issue - State: closed - Opened by sbassam almost 4 years ago - 1 comment
Labels: bug

#99 - How can I use adapters when each input has multiple [cls] tokens

Issue - State: closed - Opened by sbassam almost 4 years ago - 1 comment
Labels: question

#98 - layernorm should be set to trainable or not ?

Issue - State: closed - Opened by rabeehkarimimahabadi almost 4 years ago - 2 comments
Labels: question

#97 - is pretraining of adapter layers is necessary for performance?

Issue - State: closed - Opened by rabeehkarimimahabadi almost 4 years ago - 1 comment
Labels: question

#96 - the reason for add_classification_head

Issue - State: closed - Opened by rabeehkarimimahabadi almost 4 years ago - 2 comments
Labels: question

#95 - possible bug with adapter_bert implementation

Issue - State: closed - Opened by rabeehkarimimahabadi almost 4 years ago - 2 comments
Labels: question

#94 - moving computation inside get_adapter_preparams to the adapter forward

Issue - State: closed - Opened by rabeehkarimimahabadi almost 4 years ago - 3 comments
Labels: enhancement

#93 - large memory requirement and getting low accuracy after adding adapter layers

Issue - State: closed - Opened by rabeehk almost 4 years ago - 2 comments
Labels: Stale

#92 - adapter with seq2seq models

Issue - State: closed - Opened by rabeehkarimimahabadi almost 4 years ago - 10 comments
Labels: question

#91 - ImportError: cannot import name 'AutoModelWithHeads' from 'transformers'

Issue - State: closed - Opened by rabeehkarimimahabadi almost 4 years ago - 17 comments
Labels: bug

#90 - Adapt flex head model input to HF head model input

Pull Request - State: closed - Opened by calpt almost 4 years ago

#89 - Add Colab notebooks for adapter training, AdapterFusion & MAD-X

Pull Request - State: closed - Opened by calpt almost 4 years ago

#88 - Split Prediction Heads and make Custom Heads possible

Pull Request - State: closed - Opened by hSterz almost 4 years ago

#87 - Unintuitive slowdown in data loading and model updating on using adapters

Issue - State: closed - Opened by cs1160701 almost 4 years ago - 9 comments
Labels: bug

#86 - Bump notebook from 6.1.4 to 6.1.5 in /examples/lxmert

Pull Request - State: closed - Opened by dependabot[bot] almost 4 years ago
Labels: dependencies

#85 - Sync/v3.5.1

Pull Request - State: closed - Opened by calpt almost 4 years ago
Labels: sync

#84 - Integrate AdapterDrop

Issue - State: closed - Opened by arueckle almost 4 years ago - 3 comments
Labels: enhancement

#83 - Example code for the Inter-Adapter attention plots in Adapter Fusion

Issue - State: open - Opened by daandouwe almost 4 years ago - 2 comments
Labels: enhancement

#82 - Move master to main

Issue - State: closed - Opened by jonhilgart22 almost 4 years ago - 1 comment
Labels: enhancement

#81 - Current id2label in ModelWithHeads config

Pull Request - State: closed - Opened by hSterz almost 4 years ago - 1 comment

#80 - Support XModelWithHeads in pipelines

Pull Request - State: closed - Opened by calpt almost 4 years ago

#79 - Make sure the adapter mixin is properly implemented

Pull Request - State: closed - Opened by calpt almost 4 years ago

#78 - What is the best way to use this repo in conjunction of AllenNLP?

Issue - State: closed - Opened by arkgithubforyou almost 4 years ago - 1 comment
Labels: question

#77 - Sync/v3.4.0

Pull Request - State: closed - Opened by calpt almost 4 years ago - 1 comment
Labels: sync

#76 - Examples for zero-shot cross-lingual transfer for Classification

Issue - State: closed - Opened by PhungVanDuy almost 4 years ago - 1 comment
Labels: enhancement

#75 - Label Information

Pull Request - State: closed - Opened by hSterz almost 4 years ago

#74 - [model.load_adapter_fusion] Cannot load pre-trained adapter fusion into model

Issue - State: closed - Opened by Yujin-Yujin almost 4 years ago - 5 comments

#73 - Hinglish Sentiment Adapter

Issue - State: closed - Opened by NirantK almost 4 years ago - 8 comments
Labels: enhancement

#72 - Sync/v3.3.1

Pull Request - State: closed - Opened by calpt almost 4 years ago - 1 comment
Labels: sync

#71 - Improve docs on adapter types and configuration

Pull Request - State: closed - Opened by calpt almost 4 years ago

#70 - Updating contributing.md

Pull Request - State: closed - Opened by NirantK almost 4 years ago

#69 - Upgrade transformers from 1.0.1 to 3.x

Issue - State: closed - Opened by NirantK almost 4 years ago - 2 comments
Labels: enhancement

#68 - Fusion

Pull Request - State: closed - Opened by theorist17 almost 4 years ago

#67 - Add Adapters to DistilBERT

Pull Request - State: closed - Opened by calpt almost 4 years ago

#66 - Hotfix for adapter config saving/loading

Pull Request - State: closed - Opened by calpt almost 4 years ago
Labels: bug

#65 - Merge with original transformers library

Issue - State: closed - Opened by salimmj almost 4 years ago - 11 comments
Labels: discussion, Stale

#64 - Inference results are different every time

Issue - State: closed - Opened by z-lai almost 4 years ago - 2 comments

#63 - Add label information to adapter

Issue - State: closed - Opened by jcklie almost 4 years ago - 1 comment
Labels: enhancement

#62 - Supporting GPT-2 model

Issue - State: closed - Opened by lutz-100worte almost 4 years ago - 8 comments
Labels: enhancement

#61 - Minor fixes to Fusion code, examples, tests & docs

Pull Request - State: closed - Opened by calpt about 4 years ago

#60 - Flex head for SQuAD-style QA

Pull Request - State: closed - Opened by calpt about 4 years ago

#59 - Suggested adapter configuration for transformers with pre-layer normalization

Issue - State: closed - Opened by thevasudevgupta about 4 years ago - 2 comments
Labels: enhancement

#58 - Supporting DistilBERT model

Issue - State: closed - Opened by z-lai about 4 years ago - 1 comment
Labels: enhancement

#57 - multi-label classification / paperswithcode dataset

Issue - State: closed - Opened by ghost about 4 years ago - 2 comments
Labels: question

#56 - Loading custom adapters and 'output_attentions' for AdapterFusion

Issue - State: closed - Opened by bigkunzi about 4 years ago - 9 comments
Labels: bug, question

#55 - Sync/v3.0.2

Pull Request - State: closed - Opened by calpt about 4 years ago
Labels: sync

#54 - Cannot load saved AdapterFusion from directory with model.load_adapter_fusion()

Issue - State: closed - Opened by om304 about 4 years ago - 5 comments
Labels: bug

#53 - Language modeling head for flexible head classes

Issue - State: closed - Opened by sosuperic about 4 years ago - 6 comments
Labels: enhancement

#52 - Improvements for example scripts

Pull Request - State: closed - Opened by calpt about 4 years ago

#51 - Fix Trainer for parallel setup

Pull Request - State: closed - Opened by calpt about 4 years ago

#50 - Fix adapter args parsing and adapter saving

Pull Request - State: closed - Opened by xplip about 4 years ago

#49 - Run_squad script does not parse adapter args correctly and does not save adapters

Issue - State: closed - Opened by xplip about 4 years ago
Labels: bug

#48 - Fix leave_out adapter config attribute

Pull Request - State: closed - Opened by calpt about 4 years ago

#47 - Fix config retrieval for adapter fusion & bug in ModelAdaptersConfig

Pull Request - State: closed - Opened by calpt about 4 years ago

#46 - "leave_out" in an adapter config does not work if config is a dict

Issue - State: closed - Opened by gregor-ge about 4 years ago
Labels: bug

#45 - Updated docs for adapter quickstart & adapter training

Pull Request - State: closed - Opened by calpt about 4 years ago

#44 - Add active_adapters to all model classes with adapters

Pull Request - State: closed - Opened by calpt about 4 years ago

#43 - Quick Start Adapter was not found

Issue - State: closed - Opened by Asdf11x about 4 years ago - 2 comments
Labels: bug

#42 - Fix AdapterFusionConfig serialization

Pull Request - State: closed - Opened by calpt about 4 years ago

#41 - AdapterFusionConfig is not JSON Serializable

Issue - State: closed - Opened by arueckle about 4 years ago
Labels: bug

#40 - Added multilabel option for classification head

Pull Request - State: closed - Opened by JeroenVanHautte about 4 years ago - 2 comments

#39 - Query regarding adapter type

Issue - State: closed - Opened by prajjwal1 about 4 years ago - 6 comments

#38 - Trainer doesn't work with multi-gpu setup

Issue - State: closed - Opened by prajjwal1 about 4 years ago - 2 comments
Labels: bug

#37 - unfreezed layernorms of Bert encoders in houlsby architecture

Pull Request - State: closed - Opened by hmohebbi about 4 years ago - 1 comment

#36 - Fix loading prediction heads of task adapters

Pull Request - State: closed - Opened by calpt about 4 years ago

#35 - Run_ner.py by default tries to load prediction head from language adapter directory

Issue - State: closed - Opened by xplip about 4 years ago - 3 comments
Labels: bug

#34 - fixing issue of setting language flag in ner script

Pull Request - State: closed - Opened by JoPfeiff about 4 years ago - 1 comment

#33 - Token classification: TypeError: 'NoneType' object is not subscriptable

Issue - State: closed - Opened by stefan-it about 4 years ago - 4 comments
Labels: bug

#32 - Updates for loading from Hub

Pull Request - State: closed - Opened by calpt about 4 years ago

#31 - Rebranding readme & package setup

Pull Request - State: closed - Opened by calpt about 4 years ago - 4 comments

#30 - no warning when missing prediction head

Issue - State: closed - Opened by JoPfeiff about 4 years ago - 1 comment
Labels: bug

#29 - fix adapters with `adapter_attention`

Issue - State: closed - Opened by JoPfeiff about 4 years ago - 2 comments
Labels: bug

#28 - WIP: adapter-cli

Pull Request - State: closed - Opened by calpt about 4 years ago - 1 comment
Labels: do-not-merge

#27 - Automatically determine whether we train adapters in Trainer

Issue - State: closed - Opened by arueckle about 4 years ago - 2 comments
Labels: enhancement

#26 - Add documentation how to use loaded adapters without prediction head

Issue - State: closed - Opened by arueckle about 4 years ago - 2 comments
Labels: enhancement

#25 - Cannot load some adapters

Issue - State: closed - Opened by arueckle about 4 years ago - 2 comments
Labels: bug, wontfix

#24 - Adapter saving raises JSON serialization error

Pull Request - State: closed - Opened by arueckle about 4 years ago - 1 comment

#23 - Adapter saving raises JSON serialization errror

Issue - State: closed - Opened by arueckle about 4 years ago - 1 comment
Labels: bug

#22 - Stacking and Fusion

Pull Request - State: closed - Opened by JoPfeiff about 4 years ago - 6 comments

#21 - Fix saving of adapter prediction heads

Pull Request - State: closed - Opened by calpt about 4 years ago - 4 comments

#20 - Adapters with heads are stored multiple times after more than one checkpoint

Issue - State: closed - Opened by JoPfeiff about 4 years ago
Labels: bug

#19 - KeyError: 'mh_adapter' in add_adapter() with config containing "MH_adapter"

Issue - State: closed - Opened by gregor-ge about 4 years ago - 2 comments
Labels: bug

#18 - Fix loading of prediction heads

Pull Request - State: closed - Opened by calpt about 4 years ago

#17 - Prediction head not loaded

Issue - State: closed - Opened by arueckle about 4 years ago
Labels: bug

#16 - AdapterConfig & hub resolving updates

Pull Request - State: closed - Opened by calpt about 4 years ago

#15 - Support adapter zip folders with subfolders

Pull Request - State: closed - Opened by calpt about 4 years ago

#14 - reduction_factor -> bottleneck_size

Pull Request - State: closed - Opened by calpt about 4 years ago - 11 comments
Labels: do-not-merge

#13 - Bug fixes for adapter weights loading

Pull Request - State: closed - Opened by calpt about 4 years ago

#12 - Documentation & convenience for prediction heads

Pull Request - State: closed - Opened by calpt about 4 years ago
Labels: documentation, enhancement

#11 - fix for leave_out

Pull Request - State: closed - Opened by JoPfeiff about 4 years ago

#10 - Run tests & quality checks in CI

Pull Request - State: closed - Opened by calpt about 4 years ago

#9 - Run tests & quality checks in CI

Pull Request - State: closed - Opened by calpt about 4 years ago

#8 - Run tests & quality checks in CI

Pull Request - State: closed - Opened by calpt about 4 years ago

#7 - Remove in place additions in adapter code

Pull Request - State: closed - Opened by arueckle over 4 years ago - 1 comment

#6 - Fixed * 1.0 to .clone()

Pull Request - State: closed - Opened by JoPfeiff over 4 years ago - 1 comment

#5 - Language adapters fixes

Pull Request - State: closed - Opened by JoPfeiff over 4 years ago

#4 - Refactor adapter configs to dataclasses

Pull Request - State: closed - Opened by calpt over 4 years ago - 3 comments

#3 - Prediction Heads

Pull Request - State: closed - Opened by calpt over 4 years ago - 7 comments

#2 - Sync/v2.11.0

Pull Request - State: closed - Opened by calpt over 4 years ago - 5 comments
Labels: sync

#1 - Storing Adapters and Best Model

Pull Request - State: closed - Opened by JoPfeiff over 4 years ago - 2 comments
Labels: do-not-merge

GitHub / adapter-hub/adapters issues and pull requests