SwiftKV reduces prefill compute by up to 50% by combining model rewiring and knowledge-preserving self-distillation.
Snowflake
Enterprise
company
Verified
AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
Learn more about our AI research and open source projects, connect with our research team here.
Collections
3
A collection of text embedding models optimized for retrieval accuracy and efficiency
-
Snowflake/snowflake-arctic-embed-m
Sentence Similarity • Updated • 705k • • 150 -
Snowflake/snowflake-arctic-embed-l
Sentence Similarity • Updated • 41.4k • • 90 -
Snowflake/snowflake-arctic-embed-m-long
Sentence Similarity • Updated • 30.8k • 33 -
Snowflake/snowflake-arctic-embed-xs
Sentence Similarity • Updated • 264k • 33
models
14
Snowflake/snowflake-arctic-embed-l
Sentence Similarity
•
Updated
•
41.4k
•
•
90
Snowflake/snowflake-arctic-embed-m-v2.0
Sentence Similarity
•
Updated
•
280k
•
57
Snowflake/snowflake-arctic-embed-l-v2.0
Sentence Similarity
•
Updated
•
87k
•
119
Snowflake/snowflake-arctic-embed-m-v1.5
Sentence Similarity
•
Updated
•
32.2k
•
54
Snowflake/snowflake-arctic-embed-xs
Sentence Similarity
•
Updated
•
264k
•
33
Snowflake/snowflake-arctic-embed-m-long
Sentence Similarity
•
Updated
•
30.8k
•
33
Snowflake/snowflake-arctic-embed-m
Sentence Similarity
•
Updated
•
705k
•
•
150
Snowflake/Llama-3.1-SwiftKV-405B-Instruct-FP8
Updated
•
49
Snowflake/Llama-3.1-SwiftKV-8B-Instruct-FP8
Updated
•
165
Snowflake/Llama-3.1-SwiftKV-8B-Instruct
Updated
•
344
•
7