transfix-train/train.sh
2022-01-14 19:05:32 +01:00

18 lines
564 B
Bash
Executable File

#!/bin/bash
# arguments
# 1. root of gonito.net challenge-like filestructure
# 2. suffix of source sentences, E.g. en for files like train.en
# 3. suffix of source sentences, E.g. pl for files like train.pl
# 4. number of BPE merge operations, E.g. 32000
# 5. expected number of train epochs
corpus_path="$1"
source_suffix="$2"
target_suffix="$3"
bpe_merges="$4"
epochs="$5"
./scripts/do-fastBPE.sh "$corpus_path" "$source_suffix" "$target_suffix" "$bpe_merges"
./scripts/marian-train.sh "$corpus_path" "$source_suffix" "$target_suffix" "$bpe_merges" "$epochs"