Skip to content
Snippets Groups Projects

Repository graph

You can move around the graph by using the arrow keys.
Select Git revision
  • 823dda6dc93eb7201b35836701c879ecb9415ec0
  • master default protected
  • add_default_vectorizer_and_pretrained_loading
  • change_system_act_in_env
  • clean_code
  • convlab_exp
  • crossowoz_ddpt
  • e2e-soloist
  • emoUS
  • generalized_action_dicts
  • issue127
  • issue_114
  • nlg-scgpt
  • ppo_num_dialogues
  • pre-training
  • readme
  • remapping_actions
  • robust_masking_feature
  • scgpt_exp
  • soloist
20 results
Created with Raphaël 2.2.027Jun2623222120141310832130May27242317161312111076530Apr29272625242221201918151413111098122Mar211918171510932128Feb262523191817161514102130Jan2624211913731Dec2827222122212220191514109872130Nov2928272413Sep19Aug30Jul718Jun1720May729Apr16215Mar1428Jan29Dec18Nov131Oct30292827262218171513762130Sep2928272524232221201910987542129Aug26241916629Jul27262422181615131097429Jun23181715131025May144229Apr129Febupdate requirements.txtadd boto3 support for download in setup.py, excluding data/* using MANIFESTupdate multiwoz21 database.pyadd func relative_import_module_from_unified_datasets: import module from unified datasets, support loading module from hf datasetschange 'from data.unified_datasets.xxx import xxx' to using importlib and relative path, do not regard ConvLab-3/data as a packagemv booking_remapper into preprocess.pysupport load dataset from hf datasetuse .gitignore instead of MANIFEST.inuse Mintlify Doc Writer (vscode plugin) to generate doc for code and manually reviserename agent examplesupdate NOTICE, add new modelsupdate dockerfileupdate .gitignoreupdate github issue& pr templateupdate requirements.txt & setup.py, move tests/ to agent_examples/, move deploy to convlab/deployrename convlab2 to convlabMerge branch 'master' into clean_codeadd wow downstreampre-trainingpre-trainingadd woz datasetadd woz datasetMerge branch 'master' of gitlab.cs.uni-duesseldorf.de:dsml/convlab/ConvLab3little bugfix when using a larger amount of new dialogues before updatingMerge branch 'master' into 'master'TypeError fixedtracker gets whole history in self.staterefactored evaluate_dialog function in multiwoz evaluator. Using self.update_goal creates a mismatch of successful dialogues, where 8 dialogues have a mismatch out of 400 when using evaluate.pyMerge branch 'ddpt_ocl'can plot success for every domain combination during ocl trainingUpdate README.mdUpdate README.mdadd wow dataonline metric also saves goal so i can calculate which goals are more difficultfixed bug, goal generator now takes an additional argument defining possible goal distributionMerge branch 'master' into ddpt_oclcan use vtrace_DPT in evaluate.pyMerge branch 'github_master'option to load model weights through model_path argument in evaluate.pyupdate pre-training databug fix in plot_ocl and saving config of policy nowbug fix and added budget_balance