[refactor] move rbm.label → label.label

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

src/label/label.py

@@ -0,0 +1,119 @@
+from rbm.matrix import Mat, np
+from rbm.model import Model
+from rbm.entity import Entity, EntityParams, TrainingParams
+
+import math
+from enum import Enum
+
+class Label:
+	class EncodingType(Enum):
+		Binary = 1
+		OneHot = 2
+
+	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(), TrainingParams(num_epochs=10000, do_rao_blackwell=False))
+
+		def forward(self, x: Mat) -> Mat:
+			return self.unit1.forward(x)
+
+		def reconstruct(self, x: Mat):
+			return self.unit1.reconstruct(x)
+
+	def __init__(self, voc_size, num_dim, encoding: EncodingType, work_dir):
+		self.voc_size = voc_size
+		if encoding == Label.EncodingType.Binary:
+			self.label2vec = self.label2vec_binary
+			self.vec2label = self.vec2label_binary
+			self.fitter = Label.Fitter((round(math.log(voc_size,2)), num_dim), work_dir)
+		elif encoding == Label.EncodingType.OneHot:
+			self.label2vec = self.label2vec_onehot
+			self.vec2label = self.vec2label_onehot
+			self.fitter = Label.Fitter((voc_size, num_dim), work_dir)
+
+	def fit(self, list_of_labels: np.array):
+		label_vecs = self.label2vec(list_of_labels)
+		self.fitter.train(label_vecs)
+
+	def encode(self, list_of_labels: np.array):
+		label_vecs = self.label2vec(list_of_labels)
+		enc_data = self.fitter.forward(label_vecs)
+		return enc_data, label_vecs
+
+	def decode(self, list_of_encoded_label: np.array):
+		dec_data = self.fitter.reconstruct(list_of_encoded_label)
+		return dec_data
+
+	def label2vec_binary(self, list_of_labels: np.array):
+		num_bits = round(math.log(self.voc_size,2))
+		batch = np.zeros((len(list_of_labels), num_bits))
+		for index, label in enumerate(list_of_labels):
+			b = self._enc_binary(label, num_bits)
+			batch[index, :] = np.array(b)
+
+		return batch
+
+	def vec2label_binary(self, label_vecs: np.array, thresh=0.9):
+		labels = np.zeros((len(label_vecs)))
+		for index, label_vecs in enumerate(label_vecs):
+			labels[index] = self._dec_binary(label_vecs > thresh)
+		return labels
+
+	@staticmethod
+	def _enc_binary(number: int, num_bits):
+		res = [1 if s == '1' else 0 for s in format(number, f'{num_bits}b')]
+		return res
+
+	@staticmethod
+	def _dec_binary(vec: np.array):
+		res = 0
+		e = 1
+		for v in np.flip(vec):
+			res += v*e
+			e *= 2
+
+		return res
+
+	def label2vec_onehot(self, list_of_labels: np.array):
+		res = []
+		for label in list_of_labels:
+			vec = np.zeros(self.voc_size)
+			vec[label] = 1.0
+			res.append(vec)
+		return np.stack(res, axis=0)
+
+	def vec2label_onehot(self, label_vecs: np.array, thresh=0.9):
+		return np.argmax(label_vecs, axis=-1)
+
+if __name__ == "__main__":
+	voc_size = 100
+
+	# Create Labeler
+	enc = Label(voc_size, 16, work_dir="../../results")
+
+	# Load
+	enc.fitter.load()
+
+	# Train
+	train_labels = list(range(voc_size))
+	enc.fit(train_labels)
+
+	# Save fitter state
+	enc.fitter.save()
+
+	# Encode test labels
+	test_labels = np.array([1, 23, 99, 37, 55, 7, 31, 10, 19, 70])
+	encoded, binary_labels = enc.encode(train_labels)
+
+	# Decode
+	decoded_soft = enc.decode(encoded)
+	decode_hard = (decoded_soft > 0.9).astype(int)
+
+	# Convert soft state into labels
+	labels_reconst = enc.vec2label(decoded_soft)
+	print(labels_reconst)
+
+	print(f"Soft Error: {np.sum((decoded_soft - binary_labels) ** 2)}")
+	print(f"Hard Error: {np.sum((decode_hard - binary_labels) ** 2)}")
+	print("Test: [passed]")