updated model config

assets/best_tasks.csv

@@ -0,0 +1,105 @@
+area,task,hits
+adversarial,adversarial attack,9
+adversarial,adversarial defense,4
+adversarial,backdoor attack,1
+adversarial,adversarial text,1
+audio,sound event localization and detection,3
+audio,environmental sound classification,1
+audio,audio tagging,1
+audio,voice anti spoofing,0
+computer-code,program induction,0
+computer-code,learning to execute,0
+computer-code,type prediction,0
+computer-vision,person retrieval,10
+computer-vision,optical character recognition,10
+computer-vision,style transfer,10
+computer-vision,action recognition,10
+computer-vision,denoising,10
+computer-vision,depth estimation,10
+computer-vision,human object interaction detection,10
+computer-vision,semantic segmentation,9
+computer-vision,monocular depth estimation,8
+computer-vision,instance segmentation,8
+graphs,semantic segmentation,9
+graphs,link prediction,7
+graphs,graph attention,4
+graphs,graph clustering,3
+graphs,video inpainting,1
+graphs,graph partitioning,1
+graphs,graph regression,1
+graphs,graph reconstruction,0
+graphs,graph property prediction,0
+graphs,graph learning,0
+knowledge-base,link prediction,7
+knowledge-base,entity alignment,1
+knowledge-base,breast cancer detection,1
+medical,medical image segmentation,5
+medical,breast cancer detection,1
+medical,iris segmentation,1
+medical,cell segmentation,0
+medical,medical relation extraction,0
+medical,pancreas segmentation,0
+medical,liver segmentation,0
+medical,lesion segmentation,0
+medical,medical code prediction,0
+medical,brain segmentation,0
+methodology,gaussian processes,10
+methodology,density estimation,9
+methodology,network pruning,9
+methodology,imitation learning,9
+methodology,word embeddings,9
+methodology,outlier detection,7
+methodology,automl,6
+methodology,data augmentation,5
+methodology,stochastic optimization,5
+methodology,network embedding,4
+miscellaneous,click through rate prediction,9
+miscellaneous,session based recommendations,9
+miscellaneous,molecular property prediction,6
+miscellaneous,cross modal retrieval,5
+miscellaneous,causal inference,5
+miscellaneous,variational monte carlo,2
+miscellaneous,data visualization,0
+miscellaneous,malware classification,0
+miscellaneous,counterfactual inference,0
+miscellaneous,misinformation,0
+music,music source separation,3
+natural-language-processing,relation extraction,10
+natural-language-processing,sentiment analysis,9
+natural-language-processing,text summarization,7
+natural-language-processing,text generation,6
+natural-language-processing,aspect based sentiment analysis,5
+natural-language-processing,chinese word segmentation,4
+natural-language-processing,discourse parsing,4
+natural-language-processing,grammatical error correction,4
+natural-language-processing,discourse segmentation,3
+natural-language-processing,bilingual lexicon induction,3
+playing-games,snes games,0
+playing-games,dota 2,0
+playing-games,real time strategy games,0
+playing-games,smac,0
+reasoning,visual commonsense reasoning,2
+reasoning,model based reinforcement learning,1
+reasoning,common sense reasoning,1
+robots,semantic segmentation,9
+robots,gesture generation,2
+robots,safe exploration,0
+robots,deformable object manipulation,0
+speech,speech synthesis,10
+speech,distant speech recognition,7
+speech,speech recognition,7
+speech,speech enhancement,6
+speech,speaker diarization,3
+speech,speaker recognition,2
+speech,speaker separation,1
+speech,robust speech recognition,0
+speech,acoustic unit discovery,0
+time-series,time series,8
+time-series,dynamic time warping,6
+time-series,time series clustering,2
+time-series,stock price prediction,2
+time-series,stock market prediction,2
+time-series,video quality assessment,1
+time-series,traffic prediction,0
+time-series,time series analysis,0
+time-series,data compression,0

assets/best_tasks_with_hits.csv

@@ -1,107 +0,0 @@
-area,task,hits
-adversarial,adversarial attack,8
-adversarial,adversarial defense,2
-adversarial,data poisoning,2
-adversarial,inference attack,0
-audio,audio generation,4
-audio,audio source separation,3
-audio,language identification,1
-audio,audio tagging,1
-computer-code,text to sql,9
-computer-code,program induction,1
-computer-code,code search,0
-computer-code,type prediction,0
-computer-vision,style transfer,10
-computer-vision,crowd counting,10
-computer-vision,depth estimation,10
-computer-vision,object detection,10
-computer-vision,person re identification,10
-computer-vision,text to image generation,9
-computer-vision,image inpainting,8
-computer-vision,image registration,7
-computer-vision,video captioning,7
-computer-vision,lane detection,7
-graphs,topological data analysis,8
-graphs,node classification,7
-graphs,learning to rank,5
-graphs,graph embedding,4
-graphs,knowledge graph embedding,3
-graphs,knowledge graph embeddings,2
-graphs,graph learning,1
-graphs,knowledge base completion,1
-graphs,image outpainting,0
-graphs,image relighting,0
-knowledge-base,knowledge base completion,1
-knowledge-base,video to video synthesis,0
-knowledge-base,breast cancer detection,0
-medical,synthetic data generation,5
-medical,medical image registration,2
-medical,cardiac segmentation,2
-medical,length of stay prediction,1
-medical,skin lesion segmentation,1
-medical,medical image segmentation,1
-medical,breast cancer detection,0
-medical,liver segmentation,0
-medical,seizure detection,0
-medical,brain decoding,0
-methodology,anomaly detection,10
-methodology,active learning,10
-methodology,domain adaptation,10
-methodology,disentanglement,9
-methodology,density estimation,9
-methodology,few shot learning,7
-methodology,imitation learning,6
-methodology,stochastic optimization,6
-methodology,transfer learning,5
-methodology,outlier detection,5
-miscellaneous,survival analysis,9
-miscellaneous,recommendation systems,7
-miscellaneous,sequential recommendation,5
-miscellaneous,learning to rank,5
-miscellaneous,malware detection,5
-miscellaneous,synthetic data generation,5
-miscellaneous,multi armed bandits,4
-miscellaneous,deep clustering,3
-miscellaneous,prediction intervals,2
-miscellaneous,model discovery,1
-music,music modeling,5
-natural-language-processing,relation extraction,10
-natural-language-processing,active learning,10
-natural-language-processing,coreference resolution,10
-natural-language-processing,natural language inference,9
-natural-language-processing,text generation,8
-natural-language-processing,text matching,5
-natural-language-processing,discourse parsing,5
-natural-language-processing,relational reasoning,5
-natural-language-processing,dialogue generation,4
-natural-language-processing,fact verification,4
-playing-games,game of chess,9
-playing-games,game of go,8
-playing-games,atari games,6
-playing-games,card games,1
-playing-games,real time strategy games,1
-playing-games,starcraft,0
-playing-games,acrobot,0
-reasoning,decision making,3
-reasoning,math word problem solving,0
-robots,vision and language navigation,4
-robots,visual odometry,3
-robots,motion planning,3
-robots,image outpainting,0
-robots,robotic grasping,0
-robots,trajectory planning,0
-speech,speaker verification,9
-speech,text to speech synthesis,5
-speech,dialogue generation,4
-speech,speech enhancement,4
-speech,speaker diarization,4
-time-series,time series,8
-time-series,time series classification,3
-time-series,time series prediction,2
-time-series,multivariate time series forecasting,1
-time-series,time series clustering,1
-time-series,portfolio optimization,1
-time-series,stock market prediction,1
-time-series,activity prediction,0
-time-series,math word problem solving,0
-time-series,stock trend prediction,0

assets/worst_tasks.csv

@@ -0,0 +1,105 @@
+area,task,hits
+adversarial,backdoor attack,1
+adversarial,adversarial text,1
+adversarial,adversarial defense,4
+adversarial,adversarial attack,9
+audio,voice anti spoofing,0
+audio,environmental sound classification,1
+audio,audio tagging,1
+audio,sound event localization and detection,3
+computer-code,program induction,0
+computer-code,learning to execute,0
+computer-code,type prediction,0
+computer-vision,traffic sign recognition,0
+computer-vision,object proposal generation,0
+computer-vision,face parsing,0
+computer-vision,stereo lidar fusion,0
+computer-vision,electron microscopy image segmentation,0
+computer-vision,surface normals estimation,0
+computer-vision,salt and pepper noise removal,0
+computer-vision,weakly supervised segmentation,0
+computer-vision,camera calibration,0
+computer-vision,genre classification,0
+graphs,graph reconstruction,0
+graphs,graph property prediction,0
+graphs,graph learning,0
+graphs,video inpainting,1
+graphs,graph partitioning,1
+graphs,graph regression,1
+graphs,graph clustering,3
+graphs,graph attention,4
+graphs,link prediction,7
+graphs,semantic segmentation,9
+knowledge-base,entity alignment,1
+knowledge-base,breast cancer detection,1
+knowledge-base,link prediction,7
+medical,cell segmentation,0
+medical,medical relation extraction,0
+medical,pancreas segmentation,0
+medical,liver segmentation,0
+medical,lesion segmentation,0
+medical,medical code prediction,0
+medical,brain segmentation,0
+medical,skin cancer classification,0
+medical,medical image registration,0
+medical,breast cancer detection,1
+methodology,multi label classification,0
+methodology,unsupervised pre training,0
+methodology,classification,0
+methodology,multilingual word embeddings,0
+methodology,one class classifier,0
+methodology,data visualization,0
+methodology,automatic machine learning model selection,0
+methodology,l2 regularization,0
+methodology,abnormal event detection in video,0
+methodology,generalization bounds,0
+miscellaneous,data visualization,0
+miscellaneous,counterfactual inference,0
+miscellaneous,misinformation,0
+miscellaneous,physical simulations,0
+miscellaneous,formation energy,0
+miscellaneous,malware classification,0
+miscellaneous,multi target regression,0
+miscellaneous,open set learning,0
+miscellaneous,model discovery,0
+miscellaneous,cyber attack detection,0
+music,music source separation,3
+natural-language-processing,automatic post editing,0
+natural-language-processing,text augmentation,0
+natural-language-processing,part of speech tagging,0
+natural-language-processing,sentence ordering,0
+natural-language-processing,grammatical error detection,0
+natural-language-processing,unsupervised extractive summarization,0
+natural-language-processing,multilingual nlp,0
+natural-language-processing,text categorization,0
+natural-language-processing,twitter sentiment analysis,0
+natural-language-processing,extract aspect,0
+playing-games,snes games,0
+playing-games,dota 2,0
+playing-games,real time strategy games,0
+playing-games,smac,0
+reasoning,model based reinforcement learning,1
+reasoning,common sense reasoning,1
+reasoning,visual commonsense reasoning,2
+robots,safe exploration,0
+robots,deformable object manipulation,0
+robots,gesture generation,2
+robots,semantic segmentation,9
+speech,robust speech recognition,0
+speech,acoustic unit discovery,0
+speech,speaker separation,1
+speech,speaker recognition,2
+speech,speaker diarization,3
+speech,speech enhancement,6
+speech,distant speech recognition,7
+speech,speech recognition,7
+speech,speech synthesis,10
+time-series,traffic prediction,0
+time-series,time series analysis,0
+time-series,data compression,0
+time-series,video quality assessment,1
+time-series,time series clustering,2
+time-series,stock price prediction,2
+time-series,stock market prediction,2
+time-series,dynamic time warping,6
+time-series,time series,8

assets/worst_tasks_with_hits.csv

@@ -1,107 +0,0 @@
-area,task,hits
-adversarial,inference attack,0
-adversarial,adversarial defense,2
-adversarial,data poisoning,2
-adversarial,adversarial attack,8
-audio,language identification,1
-audio,audio tagging,1
-audio,audio source separation,3
-audio,audio generation,4
-computer-code,code search,0
-computer-code,type prediction,0
-computer-code,program induction,1
-computer-code,text to sql,9
-computer-vision,code search,0
-computer-vision,document layout analysis,0
-computer-vision,mixed reality,0
-computer-vision,stereo lidar fusion,0
-computer-vision,multi view subspace clustering,0
-computer-vision,anomaly detection in surveillance videos,0
-computer-vision,activity prediction,0
-computer-vision,superpixels,0
-computer-vision,image morphing,0
-computer-vision,hyperspectral image classification,0
-graphs,image outpainting,0
-graphs,image relighting,0
-graphs,graph reconstruction,0
-graphs,graph partitioning,0
-graphs,graph property prediction,0
-graphs,graph structure learning,0
-graphs,graph learning,1
-graphs,knowledge base completion,1
-graphs,knowledge graph embeddings,2
-graphs,knowledge graph embedding,3
-knowledge-base,video to video synthesis,0
-knowledge-base,breast cancer detection,0
-knowledge-base,knowledge base completion,1
-medical,breast cancer detection,0
-medical,liver segmentation,0
-medical,seizure detection,0
-medical,brain decoding,0
-medical,disease prediction,0
-medical,epidemiology,0
-medical,medical relation extraction,0
-medical,ecg classification,0
-medical,pancreas segmentation,0
-medical,length of stay prediction,1
-methodology,multi label classification,0
-methodology,distributed optimization,0
-methodology,dictionary learning,0
-methodology,multi label text classification,0
-methodology,generalization bounds,0
-methodology,one class classifier,0
-methodology,automatic machine learning model selection,0
-methodology,inductive logic programming,0
-methodology,bilevel optimization,0
-methodology,l2 regularization,0
-miscellaneous,fault detection,0
-miscellaneous,gpr,0
-miscellaneous,counterfactual inference,0
-miscellaneous,brain decoding,0
-miscellaneous,protein structure prediction,0
-miscellaneous,total energy,0
-miscellaneous,seismic imaging,0
-miscellaneous,vulnerability detection,0
-miscellaneous,model discovery,1
-miscellaneous,age estimation,1
-music,music modeling,5
-natural-language-processing,conversational search,0
-natural-language-processing,document ranking,0
-natural-language-processing,joint entity and relation extraction,0
-natural-language-processing,self learning,0
-natural-language-processing,intent detection,0
-natural-language-processing,text categorization,0
-natural-language-processing,opinion mining,0
-natural-language-processing,linguistic acceptability,0
-natural-language-processing,cloze test,0
-natural-language-processing,zero shot cross lingual transfer,0
-playing-games,starcraft,0
-playing-games,acrobot,0
-playing-games,card games,1
-playing-games,real time strategy games,1
-playing-games,atari games,6
-playing-games,game of go,8
-playing-games,game of chess,9
-reasoning,math word problem solving,0
-reasoning,decision making,3
-robots,image outpainting,0
-robots,robotic grasping,0
-robots,trajectory planning,0
-robots,visual odometry,3
-robots,motion planning,3
-robots,vision and language navigation,4
-speech,dialogue generation,4
-speech,speech enhancement,4
-speech,speaker diarization,4
-speech,text to speech synthesis,5
-speech,speaker verification,9
-time-series,activity prediction,0
-time-series,math word problem solving,0
-time-series,stock trend prediction,0
-time-series,change point detection,0
-time-series,spatio temporal forecasting,0
-time-series,stock price prediction,0
-time-series,multivariate time series forecasting,1
-time-series,time series clustering,1
-time-series,portfolio optimization,1
-time-series,stock market prediction,1

config.py

@@ -0,0 +1,4 @@
+query_encoder_model_name = "lambdaofgod/query_nbow_embedder"
+document_encoder_model_name = "lambdaofgod/document_nbow_embedder"
+best_tasks_path="assets/best_tasks.csv"
+worst_tasks_path="assets/worst_tasks.csv"

model_config.py

@@ -1,2 +0,0 @@
-query_encoder_model_name = "lambdaofgod/query_nbow_10_5000-15"
-document_encoder_model_name = "lambdaofgod/document_nbow_10_5000-15"

pages/1_Retrieval_App.py

@@ -7,7 +7,7 @@ import sentence_transformers
 import streamlit as st
 from findkit import feature_extractors, indexes, retrieval_pipeline
 from toolz import partial
-import model_config
+import config
 
 
 def truncate_description(description, length=50):
@@ -145,7 +145,7 @@ def app_main(
 
 
 app_main(
-    query_encoder_path=model_config.query_encoder_model_name,
-    document_encoder_path=model_config.document_encoder_model_name,
+    query_encoder_path=config.query_encoder_model_name,
+    document_encoder_path=config.document_encoder_model_name,
     data_path="lambdaofgod/pwc_repositories_with_dependencies",
 )

pages/2_Statistics.py

@@ -1,10 +1,11 @@
 import pandas as pd
 import streamlit as st
+import config
 
-best_results_df = pd.read_csv("assets/best_tasks_with_hits.csv")
+best_results_df = pd.read_csv(config.best_tasks_path)
 
 
-worst_results_df = pd.read_csv("assets/worst_tasks_with_hits.csv")
+worst_results_df = pd.read_csv(config.worst_tasks_path)
 
 show_worst_best_statistics = st.sidebar.checkbox(
     label="show worst/best statistics grouped by area"