Skip to content
New issue

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

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

Already on GitHub? Sign in to your account

Unable to reproduce MTEB leaderboard score #5

Open
AndreaSottana opened this issue Sep 20, 2024 · 0 comments
Open

Unable to reproduce MTEB leaderboard score #5

AndreaSottana opened this issue Sep 20, 2024 · 0 comments

Comments

@AndreaSottana
Copy link

AndreaSottana commented Sep 20, 2024

I am using your model Snowflake/snowflake-arctic-embed-m-v1.5 to reproduce the MTEB leaderboard results on the TRECCOVID dataset. Based on the leaderboard, the Normalized Discounted Cumulative Gain @ 10 (nDCG@10) for TRECCOVID should be 84.61. I am getting 39.39.
If you cannot see the model Snowflake/snowflake-arctic-embed-m-v1.5 easily, you should set the task to “Retrieval” only from the leaderboard and the model should be in position 35 from the top.

Below is how I tried to reproduce the results

import mteb
from sentence_transformers import SentenceTransformer

model_name = "Snowflake/snowflake-arctic-embed-m-v1.5"

model = SentenceTransformer(model_name, trust_remote_code=True)
tasks = mteb.get_tasks(tasks=["TRECCOVID"])
evaluation = mteb.MTEB(tasks=tasks)
results = evaluation.run(model, output_folder=f"results/{model_name}")
print(results[0].scores['test'][0]['ndcg_at_10'])

Expected output: 0.8641
Actual output: 0.3939

Model details: "name": "Snowflake/snowflake-arctic-embed-m-v1.5", "revision": "3b5a16eaf17e47bd997da998988dce5877a57092"

System specifications:
MacOS Sonoma 14.6.1 (23G93)
python 3.12.5
mteb 1.14.20

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

No branches or pull requests

1 participant