From 7c7097b52489e607da730bf51b8d1a03e21bc492 Mon Sep 17 00:00:00 2001
From: "lucien.noel" <noellucien2001@gmail.com>
Date: Wed, 26 Jun 2024 17:53:23 +0200
Subject: [PATCH] =?UTF-8?q?refactoring=20Transformer=20:=20g=C3=A9n=C3=A9r?=
=?UTF-8?q?er=20un=20text=20d'exemple?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---
data.py | 4 ++--
main.py | 36 +++++++++++++++++++++++++++++++-----
2 files changed, 33 insertions(+), 7 deletions(-)
diff --git a/data.py b/data.py
index 18707d8..cc80c57 100644
--- a/data.py
+++ b/data.py
@@ -33,7 +33,7 @@ class Corpus(object):
# Add words to the dictionary
with open(path, 'r', encoding="utf8") as f:
for line in f:
- words = line.split() + ['<eos>']
+ words = line.split() # + ['<eos>']
for word in words:
self.dictionary.add_word(word)
@@ -41,7 +41,7 @@ class Corpus(object):
with open(path, 'r', encoding="utf8") as f:
idss = []
for line in f:
- words = line.split() + ['<eos>']
+ words = line.split() # + ['<eos>']
ids = []
for word in words:
ids.append(self.dictionary.word2idx[word])
diff --git a/main.py b/main.py
index 5098339..02ee152 100644
--- a/main.py
+++ b/main.py
@@ -7,9 +7,9 @@ import musicgenerator as mg
device = "cuda" if torch.cuda.is_available() else "cpu"
-
if __name__ == '__main__':
learning_rate = 0.01
+ print("device :",device)
###############################################################################
# Prépare les données
@@ -28,8 +28,34 @@ if __name__ == '__main__':
###############################################################################
# Entraîne le modèle
###############################################################################
- mg.train(model, criterion, ntokens, train_data, val_data, 4, learning_rate, 5, 500)
-
-
-
+ sequence_length = 4
+ nb_log_epoch = 10
+ mg.train(
+ model=model,
+ criterion=criterion,
+ ntokens=ntokens,
+ train_data=train_data,
+ val_data=val_data,
+ sequence_length=4,
+ lr=learning_rate,
+ epochs=5,
+ log_interval=(len(train_data)//sequence_length)//nb_log_epoch
+ )
+ ###############################################################################
+ # Génère une mélodie
+ ###############################################################################
+ nb_words = 100
+ temperature = 1.0
+ model.eval()
+ input = torch.randint(ntokens, (1, 1), dtype=torch.long).to(device)
+ with torch.no_grad():
+ for i in range(nb_words):
+ output = model(input, False)
+ word_weights = output[-1].squeeze().div(temperature).exp().cpu()
+ word_idx = torch.multinomial(word_weights, 1)[0]
+ word_tensor = torch.Tensor([[word_idx]]).long().to(device)
+ input = torch.cat([input, word_tensor], 0)
+
+ word = corpus.dictionary.idx2word[word_idx]
+ print(word + ('\n' if i % 20 == 19 else ' '))
--
GitLab