Hugging Face's logo

Datasets:
gist-sparse-attention
/
GSA-PT-Qwen2-7B-Instruct-chunk8-chunk4-data

Modalities:
Tabular
Formats:
arrow
Size:
10K - 100K
Tags:
gsa
gist-sparse-attention
long-context
pretraining
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
 Community
Dataset Viewer (First 5GB)
Auto-converted to Parquet Duplicate
Search is not available for this dataset
input_ids
listlengths
2.51k
20k
labels
listlengths
2.51k
20k
length
int64
2.51k
20k
index
int64
0
617k
attention_mask
listlengths
2.51k
20k
[ 59, 2809, 90, 37155, 532, 59, 1502, 90, 5024, 14337, 299, 151646, 630, 59, 336, 759, 90, 28914, 19492, 2137, 151646, 476, 803, 3545, 1181, 6853, 33266, 11, 4221, 151646, 13026, 311, 198, 36651, 4401, 7488, 702, 1012, 151646, 151647, 263...
[ -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100...
5,900
0
[ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1...
[59,2809,90,37155,630,3966,315,279,15811,13966,304,151646,279,10126,315,2477,87289,332,1388,476,1516(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
19,558
1
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[59,2809,90,37155,630,94780,16595,320,2104,8,12378,151646,11,715,20805,438,68967,40069,320,11237,151(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
17,946
2
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[59,2809,90,37155,532,4498,15732,1276,3684,304,264,151646,40899,5476,374,65800,198,19530,50401,825,1(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
5,314
3
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[59,2809,35702,1502,90,5024,14337,299,92,37155,532,151646,715,79753,54581,315,2477,12,780,1103,15164(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
14,476
4
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[59,2809,90,37155,630,1918,279,1537,13212,11,31658,151646,44375,910,17392,51667,77413,18523,198,7,58(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
6,718
5
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[59,2809,90,37155,532,59,1502,90,5024,14337,16741,151646,532,32,67606,315,23722,614,6069,311,151646,(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
11,233
6
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[271,59,2809,90,37155,92,1124,1502,90,5024,14337,16741,151646,92,220,1124,1355,90,396,16741,532,59,2(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
17,988
7
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[59,2809,90,396,16741,532,25140,18335,315,467,1195,75,151646,5234,85266,1127,320,7433,21634,8,93,59,(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
6,974
8
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
[59,2809,90,37155,630,1109,627,278,304,947,5284,151646,572,1156,38664,311,387,315,6961,2734,151646,9(...TRUNCATED)
[-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100,-100(...TRUNCATED)
20,000
9
[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1(...TRUNCATED)
End of preview. Expand in Data Studio

GSA-PT-Qwen2-7B-Instruct-chunk8-chunk4-data

This is the continue pretraining dataset used for training GSA (Gist Sparse Attention) models based on Qwen2-7B-Instruct with chunk size chunk8-chunk4.

Each sample is tokenized and formatted with GSA gist tokens for continue pretraining.

Paper

GSA: Gist Sparse Attention via Learnable Compression and Selective Unfolding

Related Model

Downloads last month
256
Source
tokens.
Chunk size
chunk8-chunk4
Number of rows:
60,200
Estimated number of rows:
88,611
Total file size:
7.5 GB

Collection including gist-sparse-attention/GSA-PT-Qwen2-7B-Instruct-chunk8-chunk4-data