Closed tarngerine closed 1 year ago
@tarngerine Thank you for your debugging!
So please check your nodejs version:
node --version
v14.17.3
This is currently supported version (I'm working to support lts
v18)
Assumed that, let's first download an example dataset (skip this if you have yours)
cd examples/datasets/
./dataset.sh
Downloading dataset sms_dataset.tsv
Downloading dataset email_dataset.tsv
Downloading dataset cooking_dataset.tsv
cd ..
now we train a toy model (skip if you have yours):
node train
(base) musixmatch@Loretos-MBP examples % node train
train locations {
outputFile: '/var/folders/dr/lzd3czf922qcgj0xpdyg3dc40000gn/T/acc8aa68-4600-4170-a293-db0e0af20796.csv'
}
learn [
'supervised',
'-input',
'/var/folders/dr/lzd3czf922qcgj0xpdyg3dc40000gn/T/acc8aa68-4600-4170-a293-db0e0af20796.csv',
'-output',
'/Users/loreto/Documents/Projects/fasttext.js/examples/models/sms_model',
'-wordNgrams',
1,
'-minCount',
1,
...
{ W: 2810, L: 2 }
Read 0M words
Number of words: 2810
Number of labels: 2
{"progress":100,"loss":2.461702,"lr":-0.000005,"words":348044,"eta":"0h0m","eta_msec":0}
Progress: 100.0% words/sec/thread: 348044 lr: -0.000005 avg.loss: 2.461702 ETA: 0h 0m 0s {"progress":100,"loss":2.461702,"lr":0,"words":347854,"eta":"0h0m","eta_msec":0}
Progress: 100.0% words/sec/thread: 347854 lr: 0.000000 avg.loss: 2.461702 ETA: 0h 0m 0s
exec:fasttext end.
exec:fasttext exit.
task:fasttext pid:97138 terminated due to receipt of signal:null
train done. { W: 2810, L: 2 }
I have finally created a fully example to load FastText as a ES module:
cd examples/npm/
npm install
node predict.js
node predict.js
[
-0.0078989, 0.0089023, 0.0027043, -0.026109, 0.028987,
-0.027697, 0.014976, 0.012494, -0.0063605, 0.03736,
-0.0075405, 0.0027707, 0.049774, 0.00044448, 0.0017207,
-0.063726, -0.025359, 0.033446, 0.022255, -0.019012,
0.03572, 0.062952, 0.012511, -0.044282, 0.0050699,
0.0060485, 0.028198, 0.038225, -0.012945, 0.016802,
-0.018767, 0.0012059, -0.012256, -0.026341, -0.024183,
-0.041386, -0.00053842, -0.018881, 0.02069, -0.015703,
-0.013368, 0.079788, 0.017048, 0.035034, 0.021716,
-0.032098, -0.010221, -0.0033772, -0.049685, -0.010836,
-0.03294, 0.019251, -0.043315, 0.024622, -0.058188,
-0.022561, 0.0061482, -0.02715, 0.083356, -0.022334,
-0.0062796, 0.071153, -0.0041701, 0.0013033, 0.045448,
0.024724, -0.021667, -0.0073495, -0.031284, -0.046745,
-0.015068, -0.020711, 0.025491, 0.0043576, -0.0632,
-0.036824, 0.0029606, -0.057371, -0.0087149, -0.0077548,
-0.020488, 0.040209, 0.033871, -0.073204, 0.017064,
-0.034065, 0.0073495, -0.010892, -0.021408, -0.0018051,
0.048548, 0.026498, 0.00098343, -0.028395, 0.029989,
-0.060101, -0.009046, 0.024894, 0.013375, 0.036261
]
Feel free to re-open in case it does not fix.
Given a script, after installing fasttext.js with npm:
My output is
I do see that loadSentence is added to the prototype in the lib/index.js file in github but it's not in the node_modules version