Comments (2)
matt-peters
commented on July 17, 2024
Each row of the batches is treated independently by the model as one long sequence with separate states, they are not combined across rows in the batch. As the sentences are shuffled during training, the model will learn to reset its internal state at each <S> so very little state is carried over from sentence to sentence, making this a non-issue in practice. Also consider that in a real dataset, the number of sentences is typically much larger then the batch size, so repeating sentences across rows in the same batch rarely, if ever, occurs.
from bilm-tf.
tsteffek
commented on July 17, 2024
Ah, yes, I forgot that you guys shuffled the sentences. That changes things.
Thank you for your quick answer!
from bilm-tf.
Related Issues (20)
- UnicodeError when run example: python -m unittest discover tests/ HOT 2
- how many time it takes if I training a 5GB corpus on 1080ti
- How to use ELMo in Keras version? HOT 1
- why should vocabulary file be sorted in descending order by token count in our training data๏ผ
- Training ELMO on TPU for generating embedding from custom dataset
- how to freeze some layers during training HOT 1
- Import Error: No module named bilm.training when I try to incrementally train ELMo HOT 1
- drastic impact of Changing the vocabulary on perplexity HOT 1
- incremental training on less than 2000 sentences HOT 2
- elmo sentence embeddings using usage_token.py
- The PyTorch version replication of the "Deep contextualized word representations" is missing
- Does any code about ELMO+ESIM? HOT 1
- Exporting finetuned model to SavedModel format for Tensorflow Serving HOT 6
- Compute training time
- Preparing training data for a domain with many multi keyword token HOT 1
- Running into error on tests HOT 3
- Saving the custom trained ELMO embedding (hdf5) in the format of tensorhub models
- 2021-10-29 14:13:30.313623: E C:\users\nwani\_bazel_nwani\mmtm6wb6\execroot\org_tensorflow\tensorflow\stream_executor\cuda\cuda_event.cc:49] Error polling for event status: failed to query event: CUDA_ERROR_OUT_OF_MEMORY
- Pre-trained links do not work HOT 7
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. ๐๐๐
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google โค๏ธ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from bilm-tf.