Add ROADMAP.md to share our quarterly plans publicly. #3427
Conversation
|
@matt-gardner what detail would you add to the efforts to split out models into their own library? @brendan-ai2 could you link to the particular issues behind each performance improvement? |
ROADMAP.md
Outdated
| iii. | ||
|
|
||
| b. Add tooling so any user can easily measure the total number of floating-point operations | ||
| performed during training. |
There was a problem hiding this comment.
Current tracking ticket is here: #3436
I should note that this roadmap entry is more aggressive than the goal as currently written. I think that's okay in that the roadmap version is certainly where we want to end up, but it's a bit further out.
|
@matt-gardner ping: what detail would you add to the efforts to split out models into their own library? |
ROADMAP.md
Outdated
|
|
||
| a. [Semantic Parsing](https://github.com/allenai/allennlp-semparse) | ||
|
|
||
| b. [Language Modeling](https://github.com/allenai/allennlp-language-modeling) |
There was a problem hiding this comment.
I would list reading comprehension here instead, it's a lot closer to ready. The LM one is just a skeleton so far. I would add this after these lines:
We will test out this idea of splitting up the repositories with these two first,
fixing any issues with them by the end of this quarter. Around the end of
this quarter or early next quarter, we will finish the process with the rest of
the models, making repositories around language modeling, coreference
resolution, sequence tagging, parsing, etc.
There was a problem hiding this comment.
And you can add a reference to the ticket that @brendan-ai2 mentioned.
No description provided.