big improvements to readme #243
Conversation
epwalsh
requested review from
dirkgr,
AkshitaB,
ArjunSubramonian and
jacob-morrison
|
|
||
| This is an overview of the tasks supported by the AllenNLP Models library along with the corresponding components provided, organized by category. For a more comprehensive overview, see the [AllenNLP Models documentation](https://docs.allennlp.org/models/main/) or the [Paperswithcode page](https://paperswithcode.com/lib/allennlp). | ||
|
|
||
| - [**Classification**](https://github.com/allenai/allennlp-models/tree/main/allennlp_models/classification) |
There was a problem hiding this comment.
Perhaps add a description of what Classification is, like you've done for the other categories, e.g. prediction of discrete integer labels for examples
There was a problem hiding this comment.
Yeup, good call. I was being lazy 😜
README.md
Outdated
|
|
||
| - [**Coreference Resolution**](https://github.com/allenai/allennlp-models/tree/main/allennlp_models/coref) | ||
|
|
||
| Coreference resolution is the task of finding all expressions that refer to the same entity in a text. It is an important step for many higher level NLP tasks that involve natural language understanding such as document summarization, question answering, and information extraction. |
There was a problem hiding this comment.
Should we cite https://nlp.stanford.edu/projects/coref.shtml here?
README.md
Outdated
|
|
||
| - [**Multiple Choice**](https://github.com/allenai/allennlp-models/tree/main/allennlp_models/mc) | ||
|
|
||
| Multiple Choice tasks are tasks that involve selecting a correct answer out of a set of possible answers, given some input. |
There was a problem hiding this comment.
Personally confused on the distinction between multiple choice and classification
There was a problem hiding this comment.
Multiple choice is choosing between alternatives. Classification is evaluating each alternative on its own. In practice we can get human-level performance by classifying each option on its own, but I suspect that's a problem with the datasets. I wonder if a model that can actually compare possible answers against each other could do even better than what we do now.
There was a problem hiding this comment.
Just updated. Let me know if that makes sense.
While the situation is improving, model discoverability is still not great. This adds a lot of info to the README to help with that.