Skip to content

Conversation

Cocophotos
Copy link

Hi,

I've made some changes to port the parser to DyNet 2.0 (+ some other changes to make it more robust when finding the right directory for evaluation).

I've tested this version (MST & Transition-Based) on the PTB with gold & predicted PoS. I have the following results with predicted PoS:

MST:
Dev
LAS = 91.51 %
UAS = 93.97 %

Test
LAS = 91.74 %
UAS = 93.87 %

TB:
Dev
LAS = 91.52 %
UAS = 93.91 %

Test
LAS = 91.61 %
UAS = 93.78 %

Best

--
Corentin

@ftiryaki
Copy link

Hi do you have model files for bestarchybrid to run on dynet 2 ? it looks like the one we downloaded is for the old version of model.

@heejincs
Copy link

Hi, can I get the bestmodel for the MST & Transition-Based? What are the hyperparameters to train this, and how many iterations?

Thank you for the help in advance!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants