- training params are (again) attrubute of Entity

- ditched layout concept

src/rbm/entity.py

@@ -1,6 +1,48 @@
 from .state import RbmState
 from .matrix import prob, Mat
 from enum import Enum
+
+class TrainingParams:
+	def __init__(self,
+				 learning_rate: float = 0.1,
+				 momentum: float = 0.5,
+				 weight_decay: float = 0.0,
+				 num_epochs: int = 1000,
+				 num_gibbs_samples: int = 1,
+				 mini_batch_size: int = 0,
+				 do_rao_blackwell: bool = False,
+				 do_gibbs_sample_visible: bool = False,
+				 do_gibbs_sample_hidden: bool = False,
+				 do_batch_sample: bool = False
+				 ):
+		# Training parameters
+		self.learning_rate = learning_rate
+		self.momentum = momentum
+		self.weight_decay = weight_decay
+		self.num_epochs = num_epochs
+		self.num_gibbs_samples = num_gibbs_samples
+		self.mini_batch_size = mini_batch_size
+		self.do_rao_blackwell = do_rao_blackwell
+		self.do_gibbs_sample_visible = do_gibbs_sample_visible
+		self.do_gibbs_sample_hidden = do_gibbs_sample_hidden
+		self.do_batch_sample = do_batch_sample
+
+	@classmethod
+	def from_dict(cls, params: dict):
+		obj = TrainingParams()
+		obj.learning_rate = params["learningRate"]
+		obj.momentum = params["momentum"]
+		obj.weight_decay = params["weightDecay"]
+		obj.num_epochs = params["numEpochs"]
+		obj.num_gibbs_samples = params["numGibbs"]
+		obj.mini_batch_size = params["miniBatchSize"]
+		obj.do_rao_blackwell = params["doRaoBlackwell"]
+		obj.do_gibbs_sample_visible = params["gibbsDoSampleVisible"]
+		obj.do_gibbs_sample_hidden = params["gibbsDoSampleHidden"]
+		obj.do_batch_sample = params["doSampleBatch"]
+
+		return obj
+
 class EntityParams:
 	def __init__(self, do_gaussian_visible: bool = False, do_gaussian_hidden: bool = False, num_gibbs_samples: int = 1):
 		# Entity parameters
@@ -27,9 +69,10 @@ class Entity:
 		GB_RBM = "GB-RBM"
 		GG_RBM = "GG-RBM"
 
-	def __init__(self, shape: tuple[int, int], params: EntityParams):
+	def __init__(self, shape: tuple[int, int], params: EntityParams, training_params: TrainingParams|None = None):
 		self.shape = shape
 		self.params = params
+		self.training_params = training_params
 		self.state = RbmState.from_layer_params(shape)
 		self.grad = RbmState.from_layer_params(shape)
 		self.type = None

src/rbm/label.py

@@ -1,14 +1,13 @@
 from rbm.matrix import Mat, np
 from rbm.model import Model
-from rbm.entity import Entity, EntityParams
-from rbm.train import TrainingParams
+from rbm.entity import Entity, EntityParams, TrainingParams
 
 import math
 class Label:
 	class Fitter(Model):
 		def __init__(self, dim: tuple[int,int], work_dir: str = '.'):
 			super().__init__(f"label-{dim[0]}x{dim[1]}", work_dir)
-			self.unit1 = Entity(dim, EntityParams(num_gibbs_samples=1))
+			self.unit1 = Entity(dim, EntityParams(num_gibbs_samples=1), TrainingParams(num_epochs=10000, do_rao_blackwell=False))
 
 		def forward(self, x: Mat) -> Mat:
 			return self.unit1.forward(x)
@@ -36,7 +35,7 @@ class Label:
 
 	def fit(self, list_of_labels: np.array):
 		label_vecs = self.label2vec(list_of_labels)
-		self.fitter.train(label_vecs, TrainingParams(num_epochs=10000, do_rao_blackwell=False))
+		self.fitter.train(label_vecs)
 
 	def encode(self, list_of_labels: np.array):
 		label_vecs = self.label2vec(list_of_labels)

src/rbm/layer.py

@@ -1,12 +1,10 @@
-from .state import RbmState
-from .train import TrainingParams
-from .entity import Entity, EntityParams
+from .entity import Entity, EntityParams, TrainingParams
 
 class Layer:
 	def __init__(self, name: str, shape: tuple[int, int, int, int], entity_params: EntityParams, training_params: TrainingParams):
 		self.name = name
 		self.shape = shape
-		self.entity = Entity((shape[0]*shape[1]+shape[2], shape[3]), entity_params)
+		self.entity = Entity((shape[0]*shape[1]+shape[2], shape[3]), entity_params, training_params)
 		self.training_params = training_params
 
 	def init(self, std: float):

src/rbm/layout.py

@@ -1,29 +0,0 @@
-import numpy as np
-
-from .entity import Entity
-from .matrix import Mat
-from .train import train, cd_jens, TrainingParams
-from .status import Status
-
-class Horizontal:
-	def __init__(self, entities: list[Entity]):
-		self.units: list[Entity] = entities
-
-	def forward(self, x: Mat):
-		res: Mat = Mat([])
-		for unit in self.units:
-			x = unit.forward(x)
-			res = np.concat((res, x))
-
-	def reconstruct(self, x: Mat):
-		res: Mat = Mat([])
-		for unit in self.units:
-			x = unit.forward(x)
-			res = np.concat((res, x))
-
-	def train(self, batch: Mat, params: TrainingParams):
-		_batch = np.copy(batch)
-		for unit in self.units:
-			train(unit, _batch, params, Status())
-			_batch = unit.forward(_batch, num_gibbs=params.num_gibbs_samples)
-

src/rbm/model.py

@@ -2,7 +2,7 @@ import os
 from abc import ABC, abstractmethod
 
 from .entity import Entity
-from .train import train, TrainingParams, cd_jens
+from .train import train
 from .matrix import Mat, np
 from .status import Status
 
@@ -20,14 +20,12 @@ class Model(ABC):
 				obj_list.append(value)
 		return obj_list
 
-	def train(self, batch: Mat, params: TrainingParams|list[TrainingParams]):
+	def train(self, batch: Mat):
 		entities = self.objects(Entity)
-		if isinstance(params, TrainingParams):
-			params = [params]*len(entities)
 		_batch = np.copy(batch)
-		for entity, param in zip(entities, params):
-			train(entity, _batch, param, Status())
-			_batch = entity.forward(_batch, num_gibbs=param.num_gibbs_samples)
+		for entity in entities:
+			if train(entity, _batch, Status()):
+				_batch = entity.forward(_batch)
 
 	@abstractmethod
 	def forward(self, x: Mat) -> Mat:

src/rbm/stack_deep.py

@@ -10,8 +10,8 @@ class StackDeep(Stack):
 	def train(self, batch: Mat, status=Status()):
 		_batch = np.copy(batch)
 		for index, layer in enumerate(self.layers):
-			print(f"Train layer {index} for {layer.training_params.num_epochs} epochs")
-			train(layer.entity, _batch, layer.training_params, status=status)
+			print(f"Train layer {index} for {layer.entity.training_params.num_epochs} epochs")
+			train(layer.entity, _batch, status=status)
 			_batch = layer.entity.forward(_batch)
 
 	def pass_up(self, visible: Mat, from_layer_id: int = 0):

src/rbm/stack_factory.py

@@ -1,8 +1,7 @@
 import json
 from .stack import StackType
 from .layer import Layer
-from .entity import EntityParams
-from .train import TrainingParams
+from .entity import EntityParams, TrainingParams
 from .stack_deep import StackDeep
 from .stack_rnn import StackRnn
 

src/rbm/train.py

@@ -1,50 +1,9 @@
-from collections.abc import Callable
 from .matrix import sample, sample_gaussian, prob, rms_error_accu, Mat, np
 from .entity import Entity
 from .status import Status
 
-class TrainingParams:
-	def __init__(self,
-				 learning_rate: float = 0.1,
-				 momentum: float = 0.5,
-				 weight_decay: float = 0.0,
-				 num_epochs: int = 1000,
-				 num_gibbs_samples: int = 1,
-				 mini_batch_size: int = 0,
-				 do_rao_blackwell: bool = False,
-				 do_gibbs_sample_visible: bool = False,
-				 do_gibbs_sample_hidden: bool = False,
-				 do_batch_sample: bool = False
-				 ):
-		# Training parameters
-		self.learning_rate = learning_rate
-		self.momentum = momentum
-		self.weight_decay = weight_decay
-		self.num_epochs = num_epochs
-		self.num_gibbs_samples = num_gibbs_samples
-		self.mini_batch_size = mini_batch_size
-		self.do_rao_blackwell = do_rao_blackwell
-		self.do_gibbs_sample_visible = do_gibbs_sample_visible
-		self.do_gibbs_sample_hidden = do_gibbs_sample_hidden
-		self.do_batch_sample = do_batch_sample
-
-	@classmethod
-	def from_dict(cls, params: dict):
-		obj = TrainingParams()
-		obj.learning_rate = params["learningRate"]
-		obj.momentum = params["momentum"]
-		obj.weight_decay = params["weightDecay"]
-		obj.num_epochs = params["numEpochs"]
-		obj.num_gibbs_samples = params["numGibbs"]
-		obj.mini_batch_size = params["miniBatchSize"]
-		obj.do_rao_blackwell = params["doRaoBlackwell"]
-		obj.do_gibbs_sample_visible = params["gibbsDoSampleVisible"]
-		obj.do_gibbs_sample_hidden = params["gibbsDoSampleHidden"]
-		obj.do_batch_sample = params["doSampleBatch"]
-
-		return obj
-
-def cd_jens(entity: Entity, v_states: Mat, params: TrainingParams):
+def cd_jens(entity: Entity, v_states: Mat):
+	params = entity.training_params
 	v_probs = prob(v_states)
 	h_states = entity.forward(v_states)
 	h_probs = h_states
@@ -83,7 +42,8 @@ def cd_jens(entity: Entity, v_states: Mat, params: TrainingParams):
 
 	return dw, dbv, dbh
 
-def cd_binary_binary(entity: Entity, data_pos: Mat, params: TrainingParams):
+def cd_binary_binary(entity: Entity, data_pos: Mat):
+	params = entity.training_params
 	# Positive phase
 	h_probs_pos = prob(entity.h_given_v(data_pos))
 
@@ -112,7 +72,7 @@ def cd_binary_binary(entity: Entity, data_pos: Mat, params: TrainingParams):
 
 	return dw, dbv, dbh
 
-def cd_gaussian_binary(entity: Entity, data_pos: Mat, params: TrainingParams):
+def cd_gaussian_binary(entity: Entity, data_pos: Mat):
 	# Positive phase
 	h_probs_pos = entity.h_given_v(data_pos)
 
@@ -135,7 +95,7 @@ def cd_gaussian_binary(entity: Entity, data_pos: Mat, params: TrainingParams):
 
 	return dw, dbv, dbh
 
-def cd_gaussian_gaussian(entity: Entity, data_pos: Mat, params: TrainingParams):
+def cd_gaussian_gaussian(entity: Entity, data_pos: Mat):
 	# Positive phase
 	h_probs_pos = entity.h_given_v(data_pos)
 
@@ -170,7 +130,11 @@ def to_mini_batch(batch: Mat, mini_batch_size: int):
 
 	return mini_batches
 
-def train(entity: Entity, batch: Mat, params: TrainingParams, status: Status):
+def train(entity: Entity, batch: Mat, status: Status):
+	params = entity.training_params
+	if params is None:
+		return False
+
 	mini_batch_size = min(params.mini_batch_size, batch.shape[0]) if params.mini_batch_size > 0 else batch.shape[0]
 	d_progress = 100.0 / (batch.shape[0]*params.num_epochs)
 	progress = 0
@@ -191,7 +155,7 @@ def train(entity: Entity, batch: Mat, params: TrainingParams, status: Status):
 		entity.grad_zero()
 		for epochs in range(params.num_epochs):
 			# Contrastive divergence learning: calculate gradients
-			dwhv, dbv, dbh = cd_func(entity, mini_batch, params)
+			dwhv, dbv, dbh = cd_func(entity, mini_batch)
 
 			# Adjust weight and biases
 			grad = entity.grad_compute(dbv, dbh, dwhv, learning_rate=params.learning_rate/batch.shape[0], momentum=params.momentum, weight_decay=params.weight_decay)
@@ -213,4 +177,4 @@ def train(entity: Entity, batch: Mat, params: TrainingParams, status: Status):
 	status.on_change({"progress": {"value": round(progress), "unit": "%"},
 					  "err_rms_total": {"value": err_rms, "unit": ""}})
 
-	return None
+	return True