Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

feat: support seismic #552

Open
wants to merge 5 commits into
base: main
Choose a base branch
from
Open

Conversation

silver-ymz
Copy link
Member

@silver-ymz silver-ymz commented Aug 9, 2024

https://github.com/TusKANNy/seismic

benchmark on my local machine

  • datasets: Splade on Ms Marco small
  • machine: 8 vCPU, 32 GB, AMD Ryzen 7 7700, vmware Virtualization
  • index build time: 182.95s
  • index size: 692M

8 threads

  • accuracy: 91.18%
  • QPS: 10036.75

single thread

  • accuracy: 91.23%
  • QPS: 864.83

impl by the paper author

  • index build time: 204s
  • index size: 710M
  • accuracy: 91.73%
  • latency: 127 µs per query

bench params

  • n_postings = 4000
  • centroid_fraction = 0.1
  • summary_energy = 0.4
  • seismic_q_cut = 3
  • seismic_heap_factor = 0.9

Signed-off-by: Mingzhuo Yin <yinmingzhuo@gmail.com>
Signed-off-by: Mingzhuo Yin <yinmingzhuo@gmail.com>
Signed-off-by: Mingzhuo Yin <yinmingzhuo@gmail.com>
@silver-ymz
Copy link
Member Author

PTAL @VoVAllen @usamoi

Signed-off-by: Mingzhuo Yin <yinmingzhuo@gmail.com>
Signed-off-by: Mingzhuo Yin <yinmingzhuo@gmail.com>
@silver-ymz
Copy link
Member Author

In previous benchmark, seismic_heap_size is set to default value 100. Now in new benchmark, it's set to 10.

New results:

8 threads

  • accuracy: 91.18%
  • QPS: 10036.75

single thread

  • accuracy: 91.23%
  • QPS: 864.83

# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant