Reduce the config parsing codes (-360 Lines) for DS2 and make it looks cleaner. #227
Conversation
|
Should we allow more than 80 columns for this part of codes, making it look better? It will look like: Another choice is that we just forget about the column alignment and follows the normal python coding style. |
deep_speech_2/demo_server.py
Outdated
| add_arg('host_port', int, 8086, "Server's IP port.") | ||
| add_arg('host_ip', str, | ||
| 'localhost', | ||
| "Server's IP address.") |
There was a problem hiding this comment.
Should we allow more than 80 columns of codes in this part to make it look better?
deep_speech_2/demo_server.py
Outdated
| from data_utils.utils import read_manifest | ||
|
|
||
|
|
||
| def add_arg(argname, type, default, help, **kwargs): |
There was a problem hiding this comment.
Can we move down or out this function?
deep_speech_2/evaluate.py
Outdated
| add_arg('batch_size', int, 128, "Minibatch size.") | ||
| add_arg('trainer_count', int, 8, "# of Trainers (CPUs or GPUs).") | ||
| add_arg('beam_size', int, 500, "Beam search width.") | ||
| add_arg('parallels_bsearch',int, 12, "# of CPUs for beam search.") |
There was a problem hiding this comment.
Maybe we'd better keep the prefix num_ here, can use num_proc_beam_search or something like that.
| add_arg('decoder_method', str, | ||
| 'ctc_beam_search', | ||
| "Decoder method. Options: ctc_beam_search, ctc_greedy", | ||
| choices = ['ctc_beam_search', 'ctc_greedy']) |
There was a problem hiding this comment.
Should beam_search and greedy be enough?
There was a problem hiding this comment.
Add "ctc" to be more specific.
deep_speech_2/demo_server.py
Outdated
| add_arg('alpha', float, 0.36, "Coef of LM for beam search.") | ||
| add_arg('beta', float, 0.25, "Coef of WC for beam search.") | ||
| add_arg('cutoff_prob', float, 0.99, "Cutoff probability for pruning.") | ||
| add_arg('use_gru', bool, False, "Use GRUs instead of Simple RNNs.") |
deep_speech_2/evaluate.py
Outdated
| add_arg('lang_model_path', str, | ||
| 'lm/data/common_crawl_00.prune01111.trie.klm', | ||
| "Filepath for language model.") | ||
| add_arg('decoder_method', str, |
There was a problem hiding this comment.
Maybe decode_method or decoding_method would be better.
deep_speech_2/evaluate.py
Outdated
| add_arg('batch_size', int, 128, "Minibatch size.") | ||
| add_arg('trainer_count', int, 8, "# of Trainers (CPUs or GPUs).") | ||
| add_arg('beam_size', int, 500, "Beam search width.") | ||
| add_arg('parallels_bsearch',int, 12, "# of CPUs for beam search.") |
There was a problem hiding this comment.
Maybe we'd better keep the prefix num_, and can use num_proc_bsearch, num_proc_data or something like that.
xinghai-sun
changed the title
3 participants
Resolved #226