Update usage.

README.md

@@ -39,15 +39,9 @@ This model is trained for fine-grained document similarity tasks in **computer s
 
 ### How to use
 
-The `aspire-contextualsentence-multi-compsci` model can be used via the `transformers` library. 
+This model can be used via the `transformers` library, and some additional code to compute contextual sentence vectors and to make multiple matches using optimal transport.
 
-Since the model relies on contextual sentence embeddings additional code is needed to obtain cthese from a transformer model. Use it per this example usage script: [`aspire/examples/ex_aspire_consent.py`](https://github.com/allenai/aspire/blob/main/examples/ex_aspire_consent.py)
-
-Further, since the model relies on computing a document-document scores via a Wasserstein distance, use the linked class for computing these distances: [`AllPairMaskedWasserstein`](https://github.com/allenai/aspire/blob/07fdfd08698baa9e17601cc541ac9929694613b6/src/learning/facetid_models/pair_distances.py#L14)
-
-This code uses the [`geomloss`](https://www.kernel-operations.io/geomloss/api/pytorch-api.html) library for computing Wasserstein distances.
-
-(Additional example code to demo this more will be added in the coming weeks!)
+View example usage and sample document matches in the model github repo: [`examples/demo-contextualsentence-multim.ipynb`](https://github.com/allenai/aspire/blob/main/examples/demo-contextualsentence-multim.ipynb)
 
 ### Variable and metrics
 This model is evaluated on information retrieval datasets with document level queries. Performance here is reported on CSFCube (computer science/English). This is detailed on [github](https://github.com/allenai/aspire) and in our [paper](https://arxiv.org/abs/2111.08366). CSFCube presents a finer-grained query via selected sentences in a query abstract based on which a finer-grained retrieval must be made from candidate abstracts.