Comments (5)
I splits the dataset into smaller groups for a faster evaluation. You should run each command in evalFB15k-237.sh, this will generate the evaluation files.
from convkb.
I'm very sorry. In the end I still couldn't run your test code because of missing files.
I ran the evalFB15k.sh, still missing model-200.eval.4.txt
can you tell me, How does your model have hit@3 and hit@1 effects on each dataset?
Thank you very much!
from convkb.
You still have model-200.eval.0,1,2,3.txt. How many GPUs do you have? I noted in the first page that the evaluation process depends on the memory resources. When you open evalFB15k.sh, you can see:
CUDA_VISIBLE_DEVICES=1 nohup python eval.py --embedding_dim 100 --num_filters 50 --name FB15k-237 --useConstantInit True --model_name fb15k237 --num_splits 8 --testIdx 4 &
GPU id in this case is 1, I guess you have one GPU, so that's why you are missing model-200.eval.4.txt. You should change GPU id to 0.
To make sure everything running smoothly, you should run each command in evalFB15k.sh step-by-step, and check RAM-consuming in your GPU, you then know how many command you should run at the same time.
You can edit file eval.py to get Hits@1 and Hits@3 in a similar way to Hits@10 if you want.
from convkb.
Hi I found a similar problem when I was trying to run the code. In my runs/fb15k237/checkpoints directory, I only had "checkpoint model-200.data-00000-of-00001 model-200.index model-200.meta" these four files and I am confused how to generate model-200.eval.0,1,2,3.txt (txt files).
Thanks in advance!
from convkb.
You can see the commands in "evalFB15k-237.sh" that I just updated for a higher tf version. You can execute "evalFB15k-237.sh". You can also test an example command like: python eval.py --embedding_dim 100 --num_filters 50 --name FB15k-237 --useConstantInit --model_name fb15k237 --num_splits 8 --testIdx 7. The command will generate the text file model-200.eval.7.txt.
In case your machine is out of memory when executing "evalFB15k-237.sh", you should run each command in the file, and check RAM-consuming in your GPU, so you know how many command you should run at the same time. See the "evaluation metrics" section.
from convkb.
Related Issues (18)
- How to implement ConvKB on triplet classification HOT 1
- how can i get entity2vec100.init? HOT 1
- In entity2vec.init, is each line from top to bottom corresponding to the vec of entity id of 0, 1,2,3 etc.? HOT 2
- Hi, i try to implement this model myself, i use transe embeding i train myself, but i find that i can not reach the result you claim in your paper, i try many many times, help! HOT 1
- Hyperparameter settings for different datasets HOT 1
- How to use my own dataset? HOT 1
- no suitable image found. Did find: HOT 1
- How to run ConvKB in windows HOT 4
- Installation Environment
- why batch_size is so big?
- Issue try running the code HOT 3
- Interpretability of results/test HOT 3
- a problem in the eval.py HOT 11
- Training on FB15k-237 without pre-trained embeddings. HOT 1
- About python train.py
- bugs in eval.py HOT 1
- Finding validation loss HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 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. 📊📈🎉
-
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 ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from convkb.