Embedding
bekko-embedding-v1-unsupervised
Bekko Embedding v1 Unsupervised Training Data
This is the unsupervised training dataset used for the bekko-embedding-v1 embedding model family. It contains pair and triplet examples for embedding pretraining, published as Hugging Face dataset subsets so that each source/subset/task combination can be loaded independently.
For the full training recipe and technical details, see Bekko Embedding: Parameter-Efficient Multilingual Retrieval with Ultra-Compact Encoders.
The dataset… See the full description on the dataset page: https://huggingface.co/datasets/hotchpotch/bekko-embedding-v1-unsupervised.WSI_Embedding
WSI_Embedding
Patch and slide embeddings for whole-slide images, plus the run records used to produce them.
Original WSIs are not in this repo. They stay on HPC / the source dataset remotes. Do not expect .svs / .tif here.
Repo: thanminh01/WSI_Embedding
Top-level layout
Path
What it is
Download when
embeddings/<dataset>/<model>/<mag>x_<patch>px_<overlap>px_overlap/
Encoder tensors (.h5 / .pt / WSI-LLaVA json) and per-model TRIDENT _config_ / _logs_
You… See the full description on the dataset page: https://huggingface.co/datasets/thanminh01/WSI_Embedding.seamless-align-enA-viA.speaker-embedding.xlsr-2bseamless-align-enA-esA.speaker-embedding.w2vbert-600mvlm-teacher-embeddingseamless-align-enA-frA.speaker-embedding.hubert-xl