require 'htk_io'
gconf = {lrate = 0.8,
wcost = 1e-6,
momentum = 0.9,
frm_ext = 5,
rearrange = true, -- just to make the context order consistent with old TNet results, deprecated
frm_trim = 5, -- trim the first and last 5 frames, TNet just does this, deprecated
chunk_size = 1,
tr_scp = "/speechlab/users/mfy43/swb50/train_bp.scp",
cv_scp = "/speechlab/users/mfy43/swb50/train_cv.scp",
ali = {file = "/speechlab/users/mfy43/swb50/ref.mlf",
format = "map",
format_arg = "/speechlab/users/mfy43/swb50/dict",
dir = "*/",
ext = "lab"},
htk_conf = "/speechlab/users/mfy43/swb50/plp_0_d_a.conf",
initialized_param = {"/speechlab/users/mfy43/swb50/swb_init.nerv",
"/speechlab/users/mfy43/swb50/swb_global_transf.nerv"},
}
local input_size = 429
local output_size = 3001
local hidden_size = 2048
local trainer = nerv.Trainer
function trainer:make_layer_repo(param_repo)
local layer_repo = nerv.LayerRepo(
{
-- global transf
["nerv.BiasLayer"] =
{
blayer1 = {dim_in = {input_size}, dim_out = {input_size}, params = {bias = "bias0"}},
blayer2 = {dim_in = {input_size}, dim_out = {input_size}, params = {bias = "bias1"}}
},
["nerv.WindowLayer"] =
{
wlayer1 = {dim_in = {input_size}, dim_out = {input_size}, params = {window = "window0"}},
wlayer2 = {dim_in = {input_size}, dim_out = {input_size