Questions tagged [nmt]
11 questions
3
votes
2 answers
ValueError: Tokenizer class MarianTokenizer does not exist or is not currently imported
Get this error when trying to run a MarianMT-based nmt model.
Traceback (most recent call last):
File "/home/om/Desktop/Project/nmt-marionmt-api/inference.py", line 45, in
print(batch_inference(model_path="en-ar-model/Mark2",…
Om Rastogi
- 478
- 1
- 5
- 12
1
vote
0 answers
why the Other memory in JVM is growing, how to identify and make it free
I've built a spring-boot app to sniff document contents in computer.
the source: https://github.com/just226/messy-doc
after the sniffing work done, I did some native memory tracking and found that the OTHER memory swelled huge.
Other (reserved=901MB…
Eric
- 21
- 4
1
vote
0 answers
Google basic Blue score
We are working on a project to train a Auto ML models for each language supported by Google. We would like to know the basic Blue score set by Google NMT for all the supported languages. (Not only Auto ML supported languages, but also the V2, V3…
Media MissionCenter
- 11
- 2
1
vote
1 answer
How to calculate the number of epochs based on the size of training dataset, batch_size and the number of training steps?
I am new to ONMT, and am a bit confused about the number of epochs and training steps. I know the number of epochs is the number of times that we iterate through the training dataset, but how can I calculate the number of epochs that I trained my…
azLe
- 19
- 3
1
vote
1 answer
What are the differences between BLEU score and METEOR?
I am trying to understand the concept of evaluating the machine translation evaluation scores.
I understand how what BLEU score is trying to achieve. It looks into different n-grams like BLEU-1,BLEU-2, BLEU-3, BLEU-4 and try to match with the human…
Exploring
- 2,493
- 11
- 56
- 97
0
votes
1 answer
Linux OOM-Killer and G1 GC memory consumption
I have a Java application running on Liberica JDK 8 (HotSpot VM, G1 GC) on an Oracle Linux machine with 24 GB RAM. The application has -Xmx15g max heap size, utilizes it heavily (due to its load profile) and is the only process with such demands on…
Toparvion
- 799
- 2
- 9
- 19
0
votes
0 answers
How to make my NMT not translate (or remove) special characters?
I am using a pre-trained version of MarianMT.
It happens that I would like the model not to translate (or eliminate certain characters) special, since they protect me from certain information that should not be translated by the model. For example…
0
votes
0 answers
NMT Compiler OOM
spring boot service always oom killer
NMT info here
Comilper is OOM
I need advice
enter image description here
Currently, I use "- Djava.compiler=NONE" to solve the problem
The problem was solved, but the service was slower
It used to take 120s, but…
zhangyi
- 1
0
votes
0 answers
Applying meta-learning for neural machine translation
Does anyone have any idea of how to apply meta-learning for neural machine translation?
I have read some papers that consider multiple language pairs as different tasks for meta-learning. Is it possible to do it just with one pair of language? for…
Kyrsten
- 1
- 1
0
votes
0 answers
Implementing attention mechanism for neural machine translation
class Attention(nn.Module):
def __init__(self, hidden_size):
super(Attention, self).__init__()
self.hidden_size = hidden_size
# Create a two layer fully-connected network. Hint: Use nn.Sequential
# hidden_size*2…
Ani_Expo
- 1
- 1
0
votes
0 answers
Auto ML Model blue score
We are curious about the our Auto ML trained models. Even if our new trained Auto ML Model did not gain some blue score, when we use the model to translate in cat tools, does it use the Model that we trained, or it just uses the google V2 NMT?
Media MissionCenter
- 11
- 2