search

alan-turing-institute / ARC-SPICE

Sample Level, Pipeline Introduced Cumulative Errors (SPICE). Investigating methods for generalisable measurement of cumulative errors in multi-model (and multi-modal) ML pipelines.
MIT License
0 stars 0 forks source link

Translation Component #12

Closed lannelin closed 4 days ago

lannelin commented 2 weeks ago

Translating multiple languages into English.

using model that includes dropout https://huggingface.co/ybanas/autotrain-fr-en-translate-51410121895

error: COMET for document level translation error.

UQ: Using semantic density utilising MC Dropout.

J-Dymond commented 4 days ago

Added recent pull request to this issue, so this is implemented now. Could be re-opened when investigating additional models/uncertainty metrics.