File size: 12,483 Bytes
d82b09f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e4f6c78
d82b09f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fea1e8b
d82b09f
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
# -*- coding: utf-8 -*-
"""HabibiTranslator.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1lYP3XxUCWdiihU0mIejW_KCqTvy7-tz6
"""

import torch
torch.cuda.is_available()

import torch
import torch.nn as nn
import torch.optim as optim
import math
from datasets import load_dataset
import numpy as np
from collections import Counter
import gradio as gr

# Seting random seed for reproducibility
torch.manual_seed(42)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

dataset = load_dataset('Helsinki-NLP/tatoeba_mt', 'ara-eng', trust_remote_code=True)

# tokenization (word-level)
def tokenize(text):
    return text.split()

# Building vocabulary from dataset
def build_vocab(data, tokenizer, min_freq=2):
    counter = Counter()
    for example in data:
        counter.update(tokenizer(example['sourceString']))
        counter.update(tokenizer(example['targetString']))
    # Adding special tokens
    specials = ['<pad>', '<sos>', '<eos>', '<unk>']
    vocab = specials + [word for word, freq in counter.items() if freq >= min_freq]
    word2idx = {word: idx for idx, word in enumerate(vocab)}
    idx2word = {idx: word for word, idx in word2idx.items()}
    return word2idx, idx2word

# Converting text to tensor (adjusted to fit special tokens within max_len)
def text_to_tensor(text, vocab, tokenizer, max_len=52):
    tokens = tokenizer(text)[:max_len - 2]  # Reserving space for <sos> and <eos>
    tokens = ['<sos>'] + tokens + ['<eos>']
    tensor = [vocab.get(token, vocab['<unk>']) for token in tokens]
    return torch.tensor(tensor, dtype=torch.long)

train_data = dataset['validation']  # Using validation as training data for demo
test_data = dataset['test']

# Building shared vocabulary (for simplicity, using both languages in one vocab)
word2idx, idx2word = build_vocab(train_data, tokenize)

# Hyperparameters for data
max_len = 52  # Increased to account for <sos> and <eos>
batch_size = 32

train_data_list = list(train_data)  # Convert Dataset to list once
print(f"Length of train_data_list: {len(train_data_list)}")

def get_batches(data_list, batch_size, max_len=52):
    total_batches = len(data_list) // batch_size + (1 if len(data_list) % batch_size else 0)
    print(f"Total batches to process: {total_batches}")
    for i in range(0, len(data_list), batch_size):
        batch = data_list[i:i + batch_size]
        src_batch = [text_to_tensor(example['sourceString'], word2idx, tokenize, max_len) for example in batch]
        tgt_batch = [text_to_tensor(example['targetString'], word2idx, tokenize, max_len) for example in batch]
        src_batch = nn.utils.rnn.pad_sequence(src_batch, padding_value=word2idx['<pad>'], batch_first=False).to(device)
        tgt_batch = nn.utils.rnn.pad_sequence(tgt_batch, padding_value=word2idx['<pad>'], batch_first=False).to(device)
        if src_batch.size(0) > max_len:
            src_batch = src_batch[:max_len, :]
        elif src_batch.size(0) < max_len:
            padding = torch.full((max_len - src_batch.size(0), src_batch.size(1)), word2idx['<pad>'], dtype=torch.long).to(device)
            src_batch = torch.cat([src_batch, padding], dim=0)
        if tgt_batch.size(0) > max_len:
            tgt_batch = tgt_batch[:max_len, :]
        elif tgt_batch.size(0) < max_len:
            padding = torch.full((max_len - tgt_batch.size(0), tgt_batch.size(1)), word2idx['<pad>'], dtype=torch.long).to(device)
            tgt_batch = torch.cat([tgt_batch, padding], dim=0)
        src_batch = src_batch.transpose(0, 1)  # [batch_size, seq_len]
        tgt_batch = tgt_batch.transpose(0, 1)  # [batch_size, seq_len]
        yield src_batch, tgt_batch


print("Revised Chunk 1 (Seventh Iteration) completed: Dataset loaded and preprocessing debugged.")

class PositionalEncoding(nn.Module):
    def __init__(self, d_model, max_len=52):
        super().__init__()
        pe = torch.zeros(max_len, d_model)
        position = torch.arange(0, max_len, dtype=torch.float).unsqueeze(1)
        div_term = torch.exp(torch.arange(0, d_model, 2).float() * (-math.log(10000.0) / d_model))
        pe[:, 0::2] = torch.sin(position * div_term)
        pe[:, 1::2] = torch.cos(position * div_term)
        pe = pe.unsqueeze(0)  # Shape: (1, max_len, d_model)
        self.register_buffer('pe', pe)

    def forward(self, x):
        return x + self.pe[:, :x.size(1), :]

class MultiHeadAttention(nn.Module):
    def __init__(self, d_model, num_heads):
        super().__init__()
        assert d_model % num_heads == 0
        self.d_model = d_model
        self.num_heads = num_heads
        self.d_k = d_model // num_heads
        self.W_q = nn.Linear(d_model, d_model)
        self.W_k = nn.Linear(d_model, d_model)
        self.W_v = nn.Linear(d_model, d_model)
        self.W_o = nn.Linear(d_model, d_model)

    def scaled_dot_product_attention(self, Q, K, V, mask=None):
        scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(self.d_k)
        if mask is not None:
            scores = scores.masked_fill(mask == 0, -1e9)
        attn = torch.softmax(scores, dim=-1)
        return torch.matmul(attn, V)

    def forward(self, Q, K, V, mask=None):
        batch_size = Q.size(0)
        seq_len_q = Q.size(1)
        seq_len_k = K.size(1)
        Q = self.W_q(Q)
        K = self.W_k(K)
        V = self.W_v(V)
        Q = Q.view(batch_size, seq_len_q, self.num_heads, self.d_k).transpose(1, 2)
        K = K.view(batch_size, seq_len_k, self.num_heads, self.d_k).transpose(1, 2)
        V = V.view(batch_size, seq_len_k, self.num_heads, self.d_k).transpose(1, 2)
        output = self.scaled_dot_product_attention(Q, K, V, mask)
        output = output.transpose(1, 2).contiguous().view(batch_size, seq_len_q, self.d_model)
        return self.W_o(output)

class FeedForward(nn.Module):
    def __init__(self, d_model, d_ff):
        super().__init__()
        self.linear1 = nn.Linear(d_model, d_ff)
        self.linear2 = nn.Linear(d_ff, d_model)
        self.relu = nn.ReLU()

    def forward(self, x):
        return self.linear2(self.relu(self.linear1(x)))

class EncoderLayer(nn.Module):
    def __init__(self, d_model, num_heads, d_ff, dropout=0.1):
        super().__init__()
        self.mha = MultiHeadAttention(d_model, num_heads)
        self.ff = FeedForward(d_model, d_ff)
        self.norm1 = nn.LayerNorm(d_model)
        self.norm2 = nn.LayerNorm(d_model)
        self.dropout = nn.Dropout(dropout)

    def forward(self, x, mask=None):
        attn_output = self.mha(x, x, x, mask)
        x = self.norm1(x + self.dropout(attn_output))
        ff_output = self.ff(x)
        return self.norm2(x + self.dropout(ff_output))

class DecoderLayer(nn.Module):
    def __init__(self, d_model, num_heads, d_ff, dropout=0.1):
        super().__init__()
        self.mha1 = MultiHeadAttention(d_model, num_heads)
        self.mha2 = MultiHeadAttention(d_model, num_heads)
        self.ff = FeedForward(d_model, d_ff)
        self.norm1 = nn.LayerNorm(d_model)
        self.norm2 = nn.LayerNorm(d_model)
        self.norm3 = nn.LayerNorm(d_model)
        self.dropout = nn.Dropout(dropout)

    def forward(self, x, enc_output, src_mask=None, tgt_mask=None):
        attn1_output = self.mha1(x, x, x, tgt_mask)
        x = self.norm1(x + self.dropout(attn1_output))
        attn2_output = self.mha2(x, enc_output, enc_output, src_mask)
        x = self.norm2(x + self.dropout(attn2_output))
        ff_output = self.ff(x)
        return self.norm3(x + self.dropout(ff_output))

class Transformer(nn.Module):
    def __init__(self, src_vocab_size, tgt_vocab_size, d_model=256, num_heads=8, num_layers=3, d_ff=1024, max_len=52, dropout=0.1):
        super().__init__()
        self.d_model = d_model
        self.src_embedding = nn.Embedding(src_vocab_size, d_model)
        self.tgt_embedding = nn.Embedding(tgt_vocab_size, d_model)
        self.pos_encoding = PositionalEncoding(d_model, max_len)
        self.encoder_layers = nn.ModuleList([EncoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])
        self.decoder_layers = nn.ModuleList([DecoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])
        self.fc_out = nn.Linear(d_model, tgt_vocab_size)
        self.dropout = nn.Dropout(dropout)

    def generate_mask(self, src, tgt):
        src_mask = (src != word2idx['<pad>']).unsqueeze(1).unsqueeze(2)
        tgt_mask = (tgt != word2idx['<pad>']).unsqueeze(1).unsqueeze(3)
        seq_len = tgt.size(1)
        nopeak_mask = (1 - torch.triu(torch.ones(1, seq_len, seq_len), diagonal=1)).bool().to(device)
        tgt_mask = tgt_mask & nopeak_mask
        return src_mask, tgt_mask

    def forward(self, src, tgt):
        src_mask, tgt_mask = self.generate_mask(src, tgt)
        src_embedded = self.dropout(self.pos_encoding(self.src_embedding(src) * math.sqrt(self.d_model)))
        tgt_embedded = self.dropout(self.pos_encoding(self.tgt_embedding(tgt) * math.sqrt(self.d_model)))

        enc_output = src_embedded
        for enc_layer in self.encoder_layers:
            enc_output = enc_layer(enc_output, src_mask)

        dec_output = tgt_embedded
        for dec_layer in self.decoder_layers:
            dec_output = dec_layer(dec_output, enc_output, src_mask, tgt_mask)

        return self.fc_out(dec_output)

print("Revised Chunk 2 (Fourth Iteration) completed: Transformer model fixed with max_len=52.")

vocab_size = len(word2idx)
model = Transformer(
    src_vocab_size=vocab_size,
    tgt_vocab_size=vocab_size,
    d_model=256,
    num_heads=8,
    num_layers=3,
    d_ff=1024,
    max_len=52,
    dropout=0.1
).to(device)

# Loss and optimizer
criterion = nn.CrossEntropyLoss(ignore_index=word2idx['<pad>'])
optimizer = optim.Adam(model.parameters(), lr=0.0001)

# Training loop with progress feedback
def train(model, data, epochs=20):
    model.train()
    total_batches = len(data) // batch_size + (1 if len(data) % batch_size else 0)
    print(f"Total batches per epoch: {total_batches}")
    for epoch in range(epochs):
        total_loss = 0
        for batch_idx, (src_batch, tgt_batch) in enumerate(get_batches(data, batch_size, max_len=52), 1):
            if batch_idx % 100 == 0:  # Printing every 100 batches for feedback
                print(f"Epoch {epoch + 1}, Batch {batch_idx}/{total_batches} ")
            optimizer.zero_grad()
            output = model(src_batch, tgt_batch[:, :-1])
            loss = criterion(output.view(-1, vocab_size), tgt_batch[:, 1:].reshape(-1))
            loss.backward()
            optimizer.step()
            total_loss += loss.item()
        avg_loss = total_loss / total_batches
        print(f"Epoch {epoch + 1}/{epochs}, Loss: {avg_loss:.4f}")

# Main function
def translate(model, sentence, max_len=52):
    model.eval()
    with torch.no_grad():
        src = text_to_tensor(sentence, word2idx, tokenize, max_len).unsqueeze(0).to(device)
        tgt = torch.tensor([word2idx['<sos>']], dtype=torch.long).unsqueeze(0).to(device)
        for _ in range(max_len):
            output = model(src, tgt)
            next_token = output[:, -1, :].argmax(dim=-1).item()
            if next_token == word2idx['<eos>']:
                break
            tgt = torch.cat([tgt, torch.tensor([[next_token]], dtype=torch.long).to(device)], dim=1)
        translated = [idx2word[idx.item()] for idx in tgt[0] if idx.item() in idx2word]
        return ' '.join(translated[1:])


# Testing
test_sentence = "ุนู…ุฑูƒ ุฑุงูŠุญ ุงู„ู…ูƒุณูŠูƒุŸ"
translated = translate(model, test_sentence)
print(f"Input: {test_sentence}")
print(f"Translated: {translated}")

print("Chunk 3 completed: Training and inference implemented.")

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Instantiate the model (assuming train_dataset is already defined)
model = Transformer(
    src_vocab_size=vocab_size,
    tgt_vocab_size=vocab_size
).to(device)

# Load model checkpoint and set to evaluation mode
model.load_state_dict(torch.load("habibi.pth", map_location=device))
model.eval()

def gradio_translate(text):
    return translate(model, text)

interface = gr.Interface(
    fn=gradio_translate,
    inputs=gr.Textbox(lines=2, placeholder="Enter Arabic sentence here..."),
    outputs="text",
    title="Habibi-Translator",
    description="Translate Arabic sentences to English using a Transformer model."
)

interface.launch()

print("Chunk 4 completed: Gradio interface deployed.")