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

Develop Transcription->Translation->Summarisation Pipeline #1

Closed eddableheath closed 1 month ago

eddableheath commented 1 month ago

One of the two example pipelines that we want to look at to begin with has the following steps:

  1. Audio transcription.
  2. Translation.
  3. Summarisation of text.

This can be put together with models from huggingface, but some care should be given to how the error measurement can slot in here.