Hugging Face's logo

Datasets:
prompterminal
/
gpt2-compressed-weights

Dataset card Files
xet
Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot extract the features (columns) for the split 'train' of the config 'default' of the dataset.
Error code:   FeaturesError
Exception:    ArrowInvalid
Message:      Schema at index 1 was different: 
compression_ratio: string
original_parameters: int64
compressed_parameters: int64
training_method: string
base_model: string
source_weights: string
final_model: string
compression_achieved: string
status: string
vs
base_model: string
fine_tuning_method: string
batch_size: int64
accumulation_steps: int64
max_length: int64
learning_rate: double
samples_processed: int64
parameter_updates: int64
final_loss: double
compression_ratio: string
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 228, in compute_first_rows_from_streaming_response
                  iterable_dataset = iterable_dataset._resolve_features()
                File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/iterable_dataset.py", line 3422, in _resolve_features
                  features = _infer_features_from_batch(self.with_format(None)._head())
                File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/iterable_dataset.py", line 2187, in _head
                  return next(iter(self.iter(batch_size=n)))
                File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/iterable_dataset.py", line 2391, in iter
                  for key, example in iterator:
                File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/iterable_dataset.py", line 1882, in __iter__
                  for key, pa_table in self._iter_arrow():
                File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/iterable_dataset.py", line 1904, in _iter_arrow
                  yield from self.ex_iterable._iter_arrow()
                File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/iterable_dataset.py", line 559, in _iter_arrow
                  yield new_key, pa.Table.from_batches(chunks_buffer)
                File "pyarrow/table.pxi", line 4116, in pyarrow.lib.Table.from_batches
                File "pyarrow/error.pxi", line 154, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 91, in pyarrow.lib.check_status
              pyarrow.lib.ArrowInvalid: Schema at index 1 was different: 
              compression_ratio: string
              original_parameters: int64
              compressed_parameters: int64
              training_method: string
              base_model: string
              source_weights: string
              final_model: string
              compression_achieved: string
              status: string
              vs
              base_model: string
              fine_tuning_method: string
              batch_size: int64
              accumulation_steps: int64
              max_length: int64
              learning_rate: double
              samples_processed: int64
              parameter_updates: int64
              final_loss: double
              compression_ratio: string

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

GPT-2 XL Compressed Model Weights

This dataset contains the compressed model weights from tensor network compression methodology applied to GPT-2 XL.

📁 Files Included

Compressed Model Weights (.pt files)

  • compressed_gpt2_xl_68.3%.pt - Base compressed model (~68% compression)
  • compressed_gpt2_xl_68.3%_healed.pt - Compressed + knowledge distillation healing
  • compressed_gpt2_xl_68.3%_enwik8_trained.pt - Compressed + enwik8 fine-tuning
  • compressed_gpt2_xl_68.3%_enwik8_final.pt - Final version after training
  • compressed_gpt2_xl_68.3%_enwik8_finetuned.pt - Fine-tuned version

Architecture & Metadata

  • model_architecture.pkl - Compressed model architecture
  • *_metadata.json - Training and compression metadata

🔬 Methodology

Based on quantum-inspired tensor network compression:

  • Matrix Product Operator (MPO) tensor network decomposition
  • 68% parameter reduction (1.56B → ~500M parameters)
  • Tensor network compression technique
  • Knowledge distillation healing process

🚀 Usage 

import torch

# Load compressed weights
model_weights = torch.load('compressed_gpt2_xl_68.3%_healed.pt', map_location='cpu')

# For ready-to-use model, see: 
# https://huggingface.co/prompterminal/gpt2-compressed

📊 Compression Stats

  • Original GPT-2 XL: 1.56B parameters, ~6.2GB
  • Compressed Version: ~500M parameters, ~1.98GB
  • Compression Ratio: 68% reduction
  • Method: MPO tensor networks + healing

🎯 Files Recommended for Use

  • Best for inference: compressed_gpt2_xl_68.3%_healed.pt
  • Best for fine-tuning: compressed_gpt2_xl_68.3%_enwik8_trained.pt
  • Research/analysis: All files + metadata

📚 Citation 

@misc{tensor_network_compression_2024,
  title={GPT-2 XL Compressed using Tensor Network Methods},
  author={prompterminal},
  year={2024},
  howpublished={HuggingFace Dataset}
}

🔗 Related

These weights represent pioneering work in tensor network compression for large language models.

Downloads last month
2