working on lstm

1 files changed, 70 insertions, 6 deletions

diff --git a/nerv/examples/lmptb/tnn/layersT/lstm_t.lua b/nerv/examples/lmptb/tnn/layersT/lstm_t.lua
index 5b653a3..4ec2e54 100644
--- a/nerv/examples/lmptb/tnn/layersT/lstm_t.lua
+++ b/nerv/examples/lmptb/tnn/layersT/lstm_t.lua
@@ -1,6 +1,6 @@
 local LSTMLayerT = nerv.class('nerv.LSTMLayerT', 'nerv.LayerT')
 
-function LSTMLayerT:__init(id, gilobal_conf, layer_conf)
+function LSTMLayerT:__init(id, global_conf, layer_conf)
     --input1:x input2:h input3:c
     self.id = id
     self.dim_in = layer_conf.dim_in
@@ -8,13 +8,77 @@ function LSTMLayerT:__init(id, gilobal_conf, layer_conf)
     self.gconf = global_conf
 
     --prepare a DAGLayerT to hold the lstm structure
-    local paramRepo = nerv.ParamRepo()
+    local pr = layer_conf.pr
+    if pr == nil then
+        pr = nerv.ParamRepo()
+    end
+    
+    local function ap(str)
+        return self.id .. '.' .. str
+    end
+
     local layers = {
-        ["nerv.IndRecurrentLayer"] = {
-            ["recurrentL1"] = recurrentLconfig, 
-        }}
+        ["nerv.CombinerLayer"] = {
+            [ap("inputXDup")] = {{}, {["dim_in"] = {self.dim_in[1]}, ["dim_out"] = {self.dim_in[1], self.dim_in[1], self.dim_in[1]}}},
+            [ap("inputHDup")] = {{}, {["dim_in"] = {self.dim_in[2]}, ["dim_out"] = {self.dim_in[2], self.dim_in[2], self.dim_in[2]}}},
+            [ap("inputCDup")] = {{}, {["dim_in"] = {self.dim_in[3]}, 
+                ["dim_out"] = {self.dim_in[3], self.dim_in[3], self.dim_in[3], self.dim_in[3]}}},
+            [ap("mainCDup")] = {{}, {["dim_in"] = {self.dim_in[3], self.dim_in[3]}, ["dim_out"] = {self.dim_in[3], self.dim_in[3]},
+                ["lambda"] = {1, 1}}},
+        },
+        ["nerv.AffineLayer"] = {
+            [ap("mainAffineL")] = {{}, {["dim_in"] = {self.dim_in[1], self.dim_in[2], self.dim_in[3]}, 
+                ["dim_out"] = {self.dim_out[1]}, ["pr"] = pr}},
+        },
+        ["nerv.TanhLayer"] = {
+            [ap("mainTanhL")] = {{}, {["dim_in"] = {self.dim_out[1]}, ["dim_out"] = {self.dim_out[1]}}},
+            [ap("outputTanhL")] = {{}, {["dim_in"] = {self.dim_out[1]}, ["dim_out"] = {self.dim_out[1]}}},
+        },
+        ["nerv.GateFFFLayer"] = {
+            [ap("forgetGateL")] = {{}, {["dim_in"] = {self.dim_in[1], self.dim_in[2], self.dim_in[3]}, 
+                ["dim_out"] = {self.dim_in[3]}, ["pr"] = pr}},
+            [ap("inputGateL")] = {{}, {["dim_in"] = {self.dim_in[1], self.dim_in[2], self.dim_in[3]}, 
+                ["dim_out"] = {self.dim_in[3]}, ["pr"] = pr}},
+        },
+        ["nerv.ElemMulLayer"] = {
+            [ap("inputGMulL")] = {{}, {["dim_in"] = {self.dim_in[3], self.dim_in[3]}, ["dim_out"] = {self.dim_in[3]}}},
+            [ap("forgetGMulL")] = {{}, {["dim_in"] = {self.dim_in[3], self.dim_in[3]}, ["dim_out"] = {self.dim_in[3]}}},
+        },
+    }
+    
+    local layerRepo = nerv.LayerRepo(layers, pr, global_conf)
+
+    local connections_t = {
+        ["<input>[1]"] = ap("inputXDup[1]"), 
+        ["<input>[2]"] = ap("inputHDup[1]"),
+        ["<input>[3]"] = ap("inputCDup[1]"),
+
+        [ap("inputXDup[1]")] = ap("mainAffineL[1]"),
+        [ap("inputHDup[1]")] = ap("mainAffineL[2]"),
+        [ap("inputCDup[1]")] = ap("mainAffineL[3]"),
+        [ap("mainAffineL[1]")] = ap("mainTanhL[1]"),
+
+        [ap("inputXDup[2]")] = ap("inputGateL[1]"),
+        [ap("inputHDup[2]")] = ap("inputGateL[2]"),
+        [ap("inputCDup[2]")] = ap("inputGateL[3]"),
+        
+        [ap("inputXDup[3]")] = ap("forgetGateL[1]"),
+        [ap("inputHDup[3]")] = ap("forgetGateL[2]"),
+        [ap("inputCDup[3]")] = ap("forgetGateL[3]"),
+
+        [ap("mainTanhL[1]")] = ap("inputGMul[1]"),
+        [ap("inputGateL[1]")] = ap("inputGMul[2]"),
+        
+        [ap("inputCDup[4]")] = ap("forgetGMul[1]"),
+        [ap("forgetGateL[1]")] = ap("forgetGMul[2]"),
+
+        [ap("inputGMul[1]")] = ap("mainCDup[1]"),
+        [ap("forgetGMul[1]")] = ap("mainCDup[2]"),
+
+        [ap("mainCDup[2]")] = "<output>[2]",
+    }
 
-    self:check_dim_len(1, 1) -- exactly one input and one output
+    self:check_dim_len(3, 2) -- x, h, c and h, c
 end
 
 function LSTMLayerT:init(batch_size)