GEMMsTream / README.md
Efithor's picture
Update README.md
cf52ed2 verified
---
dataset_info:
features:
- name: coords
sequence:
sequence: int32
length: 3
splits:
- name: train
num_bytes: 772000000
num_examples: 1000000
download_size: 313429044
dataset_size: 772000000
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
license: mit
---
# Dataset Card for GEMMsTream
## Dataset Details
A collection 1M batches of 64 Llama 3 8B inferencing GEMMs, randomly selected. The values for N in the GEMMs were selected by selecting from the distribution of token lengths in the shareGPT dataset.