Centroid Adapter — mpnet

Lightweight BottleneckResidualAdapter trained on top of mpnet embeddings to produce representation-invariant table embeddings.

Architecture

z = e + α · Up( Dropout( GELU( Down( LN(e) ) ) ) )

Hyperparameter	Value
Embedding dim `d`	768
Bottleneck rank `r`	512
Residual scale `α`	0.01
Use bias	True

Trained on: NQ, WTQ, WIKISQL

Usage

import torch
from huggingface_hub import hf_hub_download
import json

# --- option A: use the from_pretrained helper in this repo ---
# (copy BottleneckResidualAdapter + from_pretrained from push_to_hub.py)
adapter = BottleneckResidualAdapter.from_pretrained("KBhandari11/centroid-adapter-mpnet")
e = torch.randn(1, 768)   # your backbone embedding
z = adapter(e)                    # representation-invariant embedding

# --- option B: hf_hub_download one-liner ---
from safetensors.torch import load_file
weights_path = hf_hub_download("KBhandari11/centroid-adapter-mpnet", "model.safetensors")
cfg_path     = hf_hub_download("KBhandari11/centroid-adapter-mpnet", "config.json")
with open(cfg_path) as f:
    cfg = json.load(f)
adapter = BottleneckResidualAdapter(**cfg)
adapter.load_state_dict(load_file(weights_path))
adapter.eval()