Skip to content

Running the Evaluation

This section contains documentation related to the runtime of the evaluation. How to pass arguments to the encoder, saving outputs and similar.

Simple Example

Evaluating models in mteb typically takes the simple form:

import mteb

model = mteb.get_model("sentence-transformers/static-similarity-mrl-multilingual-v1")
tasks = mteb.get_task(["MultiHateClassification"], languages = ["ita", "dan"])

results = mteb.evaluate(model, tasks=tasks)

mteb run -t MultiHateClassification -m sentence-transformers/static-similarity-mrl-multilingual-v1 -l ita dan

Compatibility with SentenceTransformers

mteb is designed to be compatible with sentence-transformers so you can also directly pass a SentenceTransformer or CrossEncoder to mteb.evaluate.

Specifying the cache

By default mteb with save the results in cache folder located at ~/.cache/mteb, however if you want to save the results in a specific folder you can specify it as follows:

cache = mteb.ResultCache(cache_path="~/.cache/mteb")
results = mteb.evaluate(model, tasks=tasks, cache=cache)

If you don't wish to run model which results already exist on the leaderboard you can download these simply by running:

cache.download_from_remote()
results = mteb.evaluate(model, tasks=tasks, cache=cache)

Tracking Carbon Emissions

mteb allows for easy tracking of carbon emission eq. using codecarbon. You simply need to install mteb[codecarbon] and enable co2 tracking:

results = mteb.evaluate(tasks=tasks, co2_tracker=True)

mteb run -t NanoArguAnaRetrieval -m sentence-transformers/static-similarity-mrl-multilingual-v1 --co2-tracker

Passing in encode arguments

To pass in arguments to the model's encode function, you can use the encode keyword arguments (encode_kwargs):

mteb.evaluate(model, tasks, encode_kwargs={"batch_size": 32})

Running SentenceTransformer model with prompts

Prompts can be passed to the SentenceTransformer model using the prompts parameter. The following code shows how to use prompts with SentenceTransformer:

from sentence_transformers import SentenceTransformer


model = SentenceTransformer("intfloat/multilingual-e5-small", prompts={"query": "Query:", "document": "Passage:"})
results = mteb.evaluate(model, tasks=tasks)

In prompts the key can be:

  1. Prompt types (passage, query) - they will be used in reranking and retrieval tasks
  2. Task type - these prompts will be used in all tasks of the given type
  3. BitextMining
  4. Classification
  5. MultilabelClassification
  6. Clustering
  7. PairClassification
  8. Reranking
  9. Retrieval
  10. STS
  11. Summarization
  12. InstructionRetrieval
  13. Pair of task type and prompt type like Retrieval-query - these prompts will be used in all Retrieval tasks
  14. Task name - these prompts will be used in the specific task
  15. Pair of task name and prompt type like NFCorpus-query - these prompts will be used in the specific task

Saving predictions

To save the predictions from a task simply set the prediction_folder:

# ...
results = mteb.evaluate(
    model,
    task,
    prediction_folder="model_predictions",
)

mteb run -t NanoArguAnaRetrieval -m sentence-transformers/static-similarity-mrl-multilingual-v1 --prediction-folder predictions

The file will now be saved to "{prediction_folder}/{task_name}_predictions.json" and contain the rankings for each query along with the model name and revision of the model that produced the result.

Speeding up Evaluations

Evaluation in MTEB consists of three main components: 1) downloading the dataset, 2) encoding of the samples, and 3) the evaluation. Typically, the most notable bottleneck are either in the encoding step or on the download step. Where this bottleneck is depends on the tasks that you evaluate on and the models that you evaluate.

If you find that any of our design decisions prevents you from running the evaluation efficiently do feel free to create an issue.

Speeding up the Model

MTEB is an evaluation framework and therefore we try to avoid doing inference optimizations. However, here is a few tricks to speed up inference.

First of all it is possible to pass directly to the model loader: 

import mteb

meta = mteb.get_model_meta("intfloat/multilingual-e5-small")

kwargs = dict(
    device="cuda", # use a gpu
    model_kwargs={"torch_dtype": "float16"}, # use low-precision
)

model = meta.load_model(**kwargs) # passed to model loader, e.g. SentenceTransformer

This e.g. allows you to use all the inference optimization tricks from sentence-transformers. However in general you can always pass device.

If you can't the required keyword argument a solution might be to extract the model: 

model = meta.load_model(**kwargs)
# extract model
sentence_trf_model = model.model

# optimizations:
sentence_trf_model.half() # half precision

A last option is to make a custom implementation of the model. This way you have full flexibility of how the model handles the input.

Speeding Download

The simplest way to speed up downloads is by using Huggingface's xet. You can use this simply using:

pip install mteb[xet]

For one of the larger datasets, MrTidyRetrieval (~15 GB), we have seen speed-ups from ~40 minutes to ~30 minutes while using xet.