Could not load model parameters in decoding

ryankiros / skip-thoughts

Sent2Vec encoder and training code from the paper "Skip-Thought Vectors"

2.05k stars 544 forks source link

Could not load model parameters in decoding #75

Open afrozas opened 6 years ago

afrozas commented 6 years ago

When we train with source sentences against target senteces

import train
train.trainer(X, C, skmodel)

the samples are seen and some form of decoding is done.

How can I use this model to decode other vectors? Essentially what is the model? I dont think it is skmodel from above, but instead some other decoding model. If yes, how can I train one or save if it is trained in this step: train.trainer(X, C, skmodel)

@ryankiros @oxingoxing Any help as you mention in this comment? Thanks!

afrozas commented 6 years ago

How much RAM should I have at hand to train decoder on ~10000 sentences? (just an estimate) On just 10 sentences, 16 GB of RAM is exhausted and the decoder is still at Packing Up step. Any help would be appreciated.