Hooshaai/bert-comp-armor-pruner-sst2

BERT-Base (armor_pruner) — NeuralOps Compressed & LoRA-Healed Model Zoo

This repository provides a physically compressed and LoRA-healed checkpoint of google-bert/bert-base-uncased fine-tuned on GLUE SST-2, compressed using the armor_pruner (ARMORPruner) engine from the NeuralOps Library at Pareto-selected rank k=256 (0.72x compression ratio, --38.8% parameter reduction).

---

Compression & Architecture Profile

ARMORPruner: Production-hardened ARMOR Pruner.

Property	Specification
Registry Key (COMPRESSOR_REGISTRY)	armor_pruner
NeuralOps Class & Family	ARMORPruner (pruning)
Source Implementation	neuralops/compression/pruning/armor_pruner.py
Base Architecture	google-bert/bert-base-uncased (109,483,778 parameters, 12 layers, $D=768$)
Showcase Rank (k)	256 (LoRA Recovery Applied: False)
Compressed Parameter Count	151,951,106 (--38.79% reduction, 0.721x physical compression ratio)
Teacher Baseline SST-2 Acc	91.63%
Showcase Final SST-2 Acc	91.40% (Zero-Shot: 91.40%, Parity Delta: -0.23%)

---

Multi-Rank Pareto Evaluation (k=256, 128, 64, 32)

Evaluated physically on NVIDIA GeForce RTX 3090 (24GB) on the GLUE SST-2 validation split (872 samples). Whenever zero-shot compression drops accuracy by > 2.0% alongside > 1.10x reduction, NeuralOps applies targeted r=8 LoRA healing (50 steps, lr=2e-4, 45s budget):

Rank (k)	Physical Ratio	Compressed Params (Reduction)	Zero-Shot Acc	Healed Acc	Parity vs Teacher (91.63%)
256 ⭐	0.72x	151,951,106 (--38.8%)	91.40%	91.40%	-0.23%
128	0.84x	130,717,442 (--19.4%)	90.83%	90.83%	-0.80%
64	0.91x	120,100,610 (--9.7%)	89.79%	89.79%	-1.84%
32	0.95x	114,792,194 (--4.8%)	88.42%	88.42%	-3.21%

(⭐ indicates the serialized checkpoint packaged in this Hugging Face repository.)

---

Packaged Artifacts

• pytorch_model.bin: Full serialized compressed BERT state dictionary (with low-rank factors / quantization buffers / LoRA healing adapters baked in).
• config.json: Model configuration tagged with compression_method: "armor_pruner" and compression_rank: 256.
• compression_summary.json: Complete machine-readable telemetry across all evaluated ranks [256, 128, 64, 32].
• Tokenizer Assets: Standard bert-base-uncased vocabulary (vocab.txt, tokenizer.json, tokenizer_config.json).

---

Quick Start (Reconstruction & Inference)

import torch
from huggingface_hub import hf_hub_download
from transformers import AutoTokenizer, BertForSequenceClassification
from neuralops.core.compression_pipeline import compress_model

repo_id = "tahamajs/bert-comp-armor-pruner-sst2"
tokenizer = AutoTokenizer.from_pretrained(repo_id)

# Reconstruct compressed NeuralOps topology and load serialized compressed + healed weights
model = BertForSequenceClassification.from_pretrained("google-bert/bert-base-uncased", num_labels=2)
model = compress_model(model, {"compressors": [{"method": "armor_pruner", "rank": 256}]})

ckpt_path = hf_hub_download(repo_id=repo_id, filename="pytorch_model.bin")
state_dict = torch.load(ckpt_path, map_location="cpu")
model.load_state_dict(state_dict, strict=False)
model.eval()

inputs = tokenizer("NeuralOps achieves strong compression without sacrificing downstream accuracy.", return_tensors="pt")
with torch.no_grad():
    logits = model(**inputs).logits
    pred = logits.argmax(dim=-1).item()
print("Predicted SST-2 Sentiment (0=Negative, 1=Positive):", pred)

---

Quick Start (Universal NeuralOps Hub API)

import torch
from neuralops.hub import load_surgeried_model

# Load model and tokenizer/processor in one line directly from Hugging Face Hub
model, tokenizer = load_surgeried_model("tahamajs/bert-comp-armor-pruner-sst2")
model.eval()
print(f"Successfully loaded tahamajs/bert-comp-armor-pruner-sst2 with NeuralOps modular surgical weights!")

Citation

@software{neuralops2026compressorzoo,
  title   = {NeuralOps: Multi-Rank Compressed BERT-SST2 Zoo with Pareto LoRA Recovery},
  author  = {Majlesi, Taha},
  year    = {2026},
  url     = {https://github.com/tahamajs/NeuralOps}
}