Skip to content

feat: evaluate arctic v2 models in MTEB(Medical)#66

Merged
KennethEnevoldsen merged 5 commits into
embeddings-benchmark:mainfrom
clinia:feat/medical-mteb-arctic-v2
Dec 11, 2024
Merged

feat: evaluate arctic v2 models in MTEB(Medical)#66
KennethEnevoldsen merged 5 commits into
embeddings-benchmark:mainfrom
clinia:feat/medical-mteb-arctic-v2

Conversation

@dbuades

@dbuades dbuades commented Dec 9, 2024

Copy link
Copy Markdown
Contributor

This PR evaluates the arctic-embed-v2.0 family of models by Snowflake on the MTEB(Medical) benchmark.

Checklist

  • Run tests locally to make sure nothing is broken using make test.
  • Run the results files checker make pre-push.

Adding a model checklist

Models are added to mteb/models in embeddings-benchmark/mteb#1574.

]
},
"evaluation_time": 33.124486207962036,
"kg_co2_emissions": null

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am a bit sad not to have the co2 usage, but otherwise, this looks fine

@dbuades dbuades Dec 10, 2024

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there any reason why codecarbon is currently not enabled by default? I'll modify my script to enable it from now on for new runs.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actually, I'll just run it again with it enabled, it doesn't take long.

@@ -0,0 +1 @@
{"name": "Snowflake/snowflake-arctic-embed-m-v2.0", "revision": "f2a7d59d80dfda5b1d14f096f3ce88bb6bf9ebdc", "release_date": null, "languages": null, "n_parameters": null, "memory_usage": null, "max_tokens": null, "embed_dim": null, "license": "apache-2.0", "open_weights": null, "public_training_data": false, "public_training_code": null, "framework": ["PyTorch"], "reference": null, "similarity_fn_name": null, "use_instructions": null, "training_datasets": null, "adapted_from": null, "superseded_by": null, "loader": null} No newline at end of file

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

a lot of the metadata seems to be missing here - is it not run using the implementation in the package? (loader: null)

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It is using the package but without embeddings-benchmark/mteb#1574. It still used the sentence_transformers_loader. I let these models ran while I was making the other PR and then once the PR was finished I just ran the models again on a single task to validate that the previous results stayed the same. I'll upload the new metadata.

@dbuades

dbuades commented Dec 10, 2024

Copy link
Copy Markdown
Contributor Author

All good @KennethEnevoldsen !

@KennethEnevoldsen KennethEnevoldsen left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Close - only a minor thing regarding use of instructions

@KennethEnevoldsen KennethEnevoldsen enabled auto-merge (squash) December 11, 2024 23:10
@KennethEnevoldsen KennethEnevoldsen merged commit edb303f into embeddings-benchmark:main Dec 11, 2024
@dbuades dbuades deleted the feat/medical-mteb-arctic-v2 branch December 13, 2024 13:18
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

Successfully merging this pull request may close these issues.

2 participants