fixed several training problems

src/rbm/cd_train.py

@@ -31,7 +31,7 @@ def cd_jens(v_states: np.ndarray, params: RbmParams, v_to_ph: Callable, h_to_pv:
 			v_probs = h_to_pv(h_probs)
 
 		# Create hidden representation given v
-		if params.do_gaussian_visible:
+		if params.do_gibbs_sample_visible:
 			h_probs = v_to_ph(sample(v_probs))
 		else:
 			h_probs = v_to_ph(v_probs)

src/rbm/layer.py

@@ -13,8 +13,8 @@ class RbmLayer:
 		self.params = params
 		self.state_filename = f"{self.name}_state.npz"
 
-	def init(self):
+	def init(self, std: float):
-		self.state.init()
+		self.state.init(mu=0, std=std)
 
 	def save(self):
 		self.state.to_file(self.state_filename)
@@ -56,7 +56,7 @@ class RbmLayer:
 				kl = self.params.learning_rate/num_cases
 				inc_bv = self.params.momentum*inc_bv + kl*dbv
 				inc_bh = self.params.momentum*inc_bh + kl*dbh
-				inc_whv = self.params.momentum*inc_whv + kl*dwhv - self.params.learning_rate*self.state.w_hv
+				inc_whv = self.params.momentum*inc_whv + kl*dwhv - self.params.weight_decay*self.state.w_hv
 
 				self.state.b_v += inc_bv
 				self.state.b_h += inc_bh
@@ -115,14 +115,14 @@ def xor():
 	params.mini_batch_size = 100
 	params.learning_rate = 0.1
 	params.momentum = 0.5
-	params.num_epochs = 100
+	params.num_epochs = 1000
 
 	status = Status()
 	layer = RbmLayer("Layer_0", 3, 16, params)
-	layer.init()
+	layer.init(0.01)
 
 	# Train
-	training_batch = np.array([[0,0,0], [0,1,1], [1,0,1], [1,1,0]], dtype=np.float64)
+	training_batch = np.array([[0,1,1], [0,0,0], [1,1,0], [1,0,1]], dtype=np.float64)
 	layer.train(training_batch, cd_jens, status)
 
 	# Test