utf8 slovenian

rm .install

type

cluster-things

sl

.compute

@@ -2,24 +2,33 @@
 
 set -xe
 
-pip3 install -r <(grep -v tensorflow requirements.txt)
-pip3 install tensorflow-gpu==1.13.0-rc2
+apt-get install -y python3-venv swig3.0
+ln -s /usr/bin/swig3.0 /usr/bin/swig
+
+python3 -m venv /tmp/venv
+source /tmp/venv/bin/activate
+
+pip install -r <(grep -v tensorflow requirements.txt)
+pip install tensorflow-gpu==1.13.0-rc2
+
+pushd native_client/ctcdecode
+make clean
+make NUM_PROCESSES=16
+pip install dist/*.whl
+popd
 
 # Install ds_ctcdecoder package from TaskCluster
-pip3 install $(python3 util/taskcluster.py --decoder)
+# pip install $(python3 util/taskcluster.py --decoder)
 
 mkdir -p ../keep/summaries
 
-data="${SHARED_DIR}/data"
-fis="${data}/LDC/fisher"
-swb="${data}/LDC/LDC97S62/swb"
-lbs="${data}/OpenSLR/LibriSpeech/librivox"
-cv="${data}/mozilla/CommonVoice/v2.0-alpha2.0/en/cv_en_valid"
+LANG="sl"
+cv="$SHARED_DIR/data/mozilla/CommonVoice/v2.0-alpha2.0/$LANG"
 
-python3 -u DeepSpeech.py \
-  --train_files '/home/josh/Downloads/sl/train.csv' \
-  --dev_files '/home/josh/Downloads/sl/dev.csv' \
-  --test_files '/home/josh/Downloads/sl/test.csv' \
+python -u DeepSpeech.py \
+  --train_files '$CV/cv_${LANG}_valid_train.csv' \
+  --dev_files '$CV/cv_${LANG}_valid_dev.csv' \
+  --test_files '$CV/cv_${LANG}_valid_test.csv' \
   --train_batch_size 24 \
   --dev_batch_size 48 \
   --test_batch_size 48 \
@@ -27,7 +36,7 @@ python3 -u DeepSpeech.py \
   --n_hidden 2048 \
   --learning_rate 0.0001 \
   --dropout_rate 0.15 \
-  --epoch 30 \
+  --epoch 1 \
   --display_step 0 \
   --validation_step 1 \
   --checkpoint_dir "../keep" \

.install

@@ -1,19 +0,0 @@
-#!/bin/bash
-
-virtualenv -p python3 ../tmp/venv
-source ../tmp/venv/bin/activate
-pip install -r <(grep -v tensorflow requirements.txt)
-pip install tensorflow
-
-python3 util/taskcluster.py --arch cpu --target ../tmp/native_client
-
-# Install ds_ctcdecoder package from TaskCluster
-# pip install $(python3 util/taskcluster.py --decoder)
-
-pushd native_client/ctcdecode
-make clean
-make NUM_PROCESSES=16
-pip install dist/*.whl
-popd
-
-mkdir -p ../keep/summaries

data/alphabet.txt

@@ -0,0 +1,64 @@
+## The following unique characters were found in your transcripts: ###
+?
+i
+V
+Š
+Z
+j
+;
+L
+z
+.
+Č
+r
+–
+o
+P
+H
+J
+ 
+,
+b
+N
+R
+u
+ž
+:
+!
+Ž
+a
+p
+s
+G
+l
+š
+B
+F
+f
+c
+T
+M
+-
+v
+W
+e
+S
+y
+m
+t
+A
+g
+4
+I
+n
+U
+E
+C
+D
+K
+č
+k
+h
+O
+d
+##