- -model  Path to model .pt file
- -data_type [text] Type of the source input. Options: [text|img].
- -src  Source sequence to decode (one line per sequence)
- -src_dir  Source directory for image or audio files
- -tgt  True target sequence (optional)
- -output [pred.txt] Path to output the predictions (each line will be the decoded sequence
- -report_bleu  Report bleu score after translation, call tools/multi-bleu.perl on command line
- -report_rouge  Report rouge 1/2/3/L/SU4 score after translation call tools/test_rouge.py on command line
- -dynamic_dict  Create dynamic dictionaries
- -share_vocab  Share source and target vocabulary
- -beam_size  Beam size
- -min_length  Minimum prediction length
- -max_length  Maximum prediction length.
- -max_sent_length 
- -stepwise_penalty  Apply penalty at every decoding step. Helpful for summary penalty.
- -length_penalty [none] Length Penalty to use.
- -coverage_penalty [none] Coverage Penalty to use.
- -alpha  Google NMT length penalty parameter (higher = longer generation)
- -beta  Coverage penalty parameter
- -block_ngram_repeat  Block repetition of ngrams during decoding.
- -ignore_when_blocking  Ignore these strings when blocking repeats. You want to block sentence delimiters.
- -replace_unk  Replace the generated UNK tokens with the source token that had highest attention weight. If phrase_table is provided, it will lookup the identified source token and give the corresponding target token. If it is not provided(or the identified source token does not exist in the table) then it will copy the source token
- -verbose  Print scores and predictions for each sentence
- -attn_debug  Print best attn for each word
- -dump_beam  File to dump beam information to.
- -n_best  If verbose is set, will output the n_best decoded sentences
- -batch_size  Batch size
- -gpu [-1] Device to run on
- -sample_rate  Sample rate.
- -window_size [0.02] Window size for spectrogram in seconds
- -window_stride [0.01] Window stride for spectrogram in seconds
- -window [hamming] Window type for spectrogram generation