Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 22 additions & 0 deletions src/dataflex/configs/components.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -40,6 +40,28 @@ selectors:
seed: 123
save_interval: 16

cluster_less:
name: cluster_less
params:
cache_dir: ../dataflex_saves/cluster_less_output
gradient_type: adam
proj_dim: 4096
seed: 42
save_interval: 16
cluster_size: 64
samples_per_cluster: 4
clustering_batch_size: 8
clustering_max_iter: 10
clustering_method: kmeans
representative_strategy: random
# Optional: standalone model for the ONE-TIME clustering embeddings.
# When set, clustering is done once with this model and frozen for the
# whole run; the training model is only used for the per-step gradient
# refresh. Leave null to use the training model (legacy behaviour).
embedding_model_name_or_path: null
embedding_model_dtype: auto
embedding_layer: -1

loss:
name: loss
params:
Expand Down
3 changes: 2 additions & 1 deletion src/dataflex/train/selector/__init__.py
Original file line number Diff line number Diff line change
Expand Up @@ -5,4 +5,5 @@
from .tsds_selector import *
from .nice_selector import *
from .random_selector import *
from .near_selector import *
from .near_selector import *
from .cluster_less_selector import *
Loading
Loading