lindsey98

Large embedding vector, large number of classes?

1

Hi authors, Thanks for pushing this fantastic tool. I have a question regarding to the approach itself: When my my last layer dimension is really large + I have large...

Return ALL the indices falling into the nearest IVF cell

3

Hi, may I ask: Given a query vector, how do I return ALL the indices falling into the nearest IVF cell?

Cuda 12 support

2

Hi, does Uni-Core currently support CUDA 12? Thanks!

replace old target-list with new target-list

2

Recently, we have expanded our targetlist to include 2K+ targets. The new expand_targetlist.zip: https://drive.google.com/file/d/105qPU_MbvPH0Y2X6YiDvOYc5pnDeGv1R/view?usp=share_link The new domain_map.pkl: https://drive.google.com/file/d/1DeoI1pjkEcPWDNAO6kTlLin0UIUZfF1Y/view?usp=share_link Step 1: Find where the PhishIntention package is installed, e.g. /anaconda3/envs/.../lib/python3.7/site-packages/phishintention Step...

TODO

During training, we use the logo images to train a 277-brand classification task. During testing, we use the second last intermediate layer as the logo embedding, and discard the classification...

All the training scripts are inside phishintention/src/*_utils, the training framework for object detector is [Detectron2](https://github.com/facebookresearch/detectron2), and for classification is [Big-transfer](https://github.com/google-research/big_transfer). 

The code for ROC plot (Figure 11) is as follows: ``` import matplotlib.pyplot as plt import numpy as np import pandas as pd from matplotlib import rcParams rcParams['font.family'] = ['serif']...

Please change the api_key to your own token [register here](https://www.virustotal.com/gui/home/upload) to avoid receiving response 403 (exceeds daily query limits). 

Please run ```bash pip install helium==3.0.9 ```