Readding documents necessary for Streamlit Document

.gitignore

@@ -0,0 +1,15 @@
+# Python
+*.pyc
+__pycache__/
+venv/
+*.env
+
+# VS Code
+.vscode/
+.settings/
+.cache/
+
+# Other
+*.log
+*.swp
+.DS_Store

README.md

@@ -1,7 +1,7 @@
 ---
 title: YorubaCNN
-emoji: 📚
-colorFrom: yellow
+emoji: 📉
+colorFrom: green
 colorTo: gray
 sdk: streamlit
 sdk_version: 1.37.1

main.py

@@ -0,0 +1 @@
+print('Hello, Lightning World!')

model_loader.py

@@ -0,0 +1,37 @@
+from huggingface_hub import HfApi, HfFolder
+import os
+import os
+from dotenv import load_dotenv
+
+# Load environment variables from .env file if it exists
+load_dotenv()
+
+hf_token = os.getenv('HF_TOKEN')
+api = HfApi()
+
+# Specify the model details
+model_id = "Testys/cnn_yor_ner"
+sent_id = "Testys/cnn_sent_yor"
+local_dir = "./my_model"
+sent_dir = "./sent_model"
+
+# Download the model folder
+if not os.path.exists(local_dir):
+    os.makedirs(local_dir)
+
+    if not os.path.exists(os.path.join(local_dir, "pytorch_model.bin")):
+        api.hf_hub_download(repo_id=model_id, filename="pytorch_model.bin", local_dir=local_dir, use_auth_token=hf_token)
+    
+    if not os.path.exists(os.path.join(local_dir, "config.json")):
+        api.hf_hub_download(repo_id=model_id, filename="config.json", local_dir=local_dir, use_auth_token=hf_token)
+
+
+# Check if the model is already downloaded
+if not os.path.exists(sent_dir):
+    os.makedirs(sent_dir) 
+    
+    # Download the model files only if they don't exist
+    if not os.path.exists(os.path.join(sent_dir, "sent_pytorch_model.bin")):
+        api.hf_hub_download(repo_id=sent_id, filename="sent_pytorch_model.bin", local_dir=sent_dir, use_auth_token=hf_token)
+    if not os.path.exists(os.path.join(sent_dir, "config.json")):
+        api.hf_hub_download(repo_id=sent_id, filename="config.json", local_dir=sent_dir, use_auth_token=hf_token)

modelling_cnn.py

@@ -0,0 +1,126 @@
+import os
+import json
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from transformers import pipeline
+from huggingface_hub import HfApi, HfFolder
+from transformers import AutoTokenizer, AutoConfig, AutoModelForTokenClassification, AutoModel
+
+
+class CNNForNER(nn.Module):
+    def __init__(self, pretrained_model_name, num_classes, max_length=128):
+        super(CNNForNER, self).__init__()
+        self.transformer = AutoModelForTokenClassification.from_pretrained(pretrained_model_name)
+        self.max_length = max_length
+
+        # Get the number of labels from the pretrained model
+        pretrained_num_labels = self.transformer.num_labels
+
+        self.conv1 = nn.Conv1d(in_channels=pretrained_num_labels, out_channels=256, kernel_size=3, padding=1)
+        self.conv2 = nn.Conv1d(in_channels=256, out_channels=128, kernel_size=3, padding=1)
+        self.dropout = nn.Dropout(0.3)
+        self.fc = nn.Linear(in_features=128, out_features=num_classes)
+
+    def forward(self, input_ids, attention_mask):
+        outputs = self.transformer(input_ids=input_ids, attention_mask=attention_mask)
+        logits = outputs.logits  # Shape: (batch_size, sequence_length, pretrained_num_labels)
+
+        # Apply CNN layers
+        logits = logits.permute(0, 2, 1)  # Shape: (batch_size, pretrained_num_labels, sequence_length)
+        conv1_out = F.relu(self.conv1(logits))
+        conv2_out = F.relu(self.conv2(conv1_out))
+        conv2_out = self.dropout(conv2_out)
+        conv2_out = conv2_out.permute(0, 2, 1)  # Shape: (batch_size, sequence_length, 128)
+        final_logits = self.fc(conv2_out)  # Shape: (batch_size, sequence_length, num_classes)
+        return final_logits
+
+
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from transformers import AutoModel
+
+class SentimentCNNModel(nn.Module):
+    def __init__(self, transformer_model_name, num_classes, cnn_out_channels=100, cnn_kernel_sizes=[3, 5, 7]):
+        super(SentimentCNNModel, self).__init__()
+        # Load pre-trained transformer model
+        self.transformer = AutoModel.from_pretrained(transformer_model_name)
+
+        # CNN layers with multiple kernel sizes
+        self.convs = nn.ModuleList([
+            nn.Conv1d(in_channels=self.transformer.config.hidden_size,
+                      out_channels=cnn_out_channels,
+                      kernel_size=k)
+            for k in cnn_kernel_sizes
+        ])
+
+        # Dropout layer
+        self.dropout = nn.Dropout(0.5)
+
+        # Fully connected layer
+        self.fc = nn.Linear(len(cnn_kernel_sizes) * cnn_out_channels, num_classes)
+
+    def forward(self, input_ids, attention_mask):
+        # Get hidden states from the transformer model
+        transformer_outputs = self.transformer(input_ids=input_ids, attention_mask=attention_mask)
+        hidden_states = transformer_outputs.last_hidden_state  # Shape: (batch_size, seq_len, hidden_size)
+
+        # Transpose for CNN input: (batch_size, hidden_size, seq_len)
+        hidden_states = hidden_states.transpose(1, 2)
+
+        # Apply convolution and pooling
+        conv_outputs = [torch.relu(conv(hidden_states)) for conv in self.convs]
+        pooled_outputs = [torch.max(output, dim=2)[0] for output in conv_outputs]
+
+        # Concatenate pooled outputs and apply dropout
+        cat_output = torch.cat(pooled_outputs, dim=1)
+        cat_output = self.dropout(cat_output)
+
+        # Final classification
+        logits = self.fc(cat_output)
+
+        return logits
+
+
+class SentimentCNNModel(nn.Module):
+    def __init__(self, transformer_model_name, num_classes, cnn_out_channels=100, cnn_kernel_sizes=[3, 5, 7]):
+        super(SentimentCNNModel, self).__init__()
+        # Load pre-trained transformer model
+        self.transformer = AutoModel.from_pretrained(transformer_model_name)
+
+        # CNN layers with multiple kernel sizes
+        self.convs = nn.ModuleList([
+            nn.Conv1d(in_channels=self.transformer.config.hidden_size,
+                      out_channels=cnn_out_channels,
+                      kernel_size=k)
+            for k in cnn_kernel_sizes
+        ])
+
+        # Dropout layer
+        self.dropout = nn.Dropout(0.5)
+
+        # Fully connected layer
+        self.fc = nn.Linear(len(cnn_kernel_sizes) * cnn_out_channels, num_classes)
+
+    def forward(self, input_ids, attention_mask):
+        # Get hidden states from the transformer model
+        transformer_outputs = self.transformer(input_ids=input_ids, attention_mask=attention_mask)
+        hidden_states = transformer_outputs.last_hidden_state  # Shape: (batch_size, seq_len, hidden_size)
+
+        # Transpose for CNN input: (batch_size, hidden_size, seq_len)
+        hidden_states = hidden_states.transpose(1, 2)
+
+        # Apply convolution and pooling
+        conv_outputs = [torch.relu(conv(hidden_states)) for conv in self.convs]
+        pooled_outputs = [torch.max(output, dim=2)[0] for output in conv_outputs]
+
+        # Concatenate pooled outputs and apply dropout
+        cat_output = torch.cat(pooled_outputs, dim=1)
+        cat_output = self.dropout(cat_output)
+
+        # Final classification
+        logits = self.fc(cat_output)
+
+        return logits

my_model/config.json

@@ -0,0 +1 @@
+{"model_type": "CNNForYorubaNER", "num_classes": 9, "max_length": 128, "pretrained_model_name": "masakhane/afroxlmr-large-ner-masakhaner-1.0_2.0"}

my_model/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6df5d161a93b1a7470e4e0cb4f1d8a9dbeb74a9fb56b72d318dad0731c9a379
+size 2236003390

requirements.txt

@@ -0,0 +1,4 @@
+torch
+transformers
+huggingface_hub
+streamlit

sent_model/config.json

@@ -0,0 +1 @@
+{"model_type": "CNNForSentimentAnalysis", "num_classes": 2, "max_length": 128, "pretrained_model_name": "Davlan/naija-twitter-sentiment-afriberta-large"}

sent_model/sent_pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f9eb93cf4592a046678a953616e4281e7bba91703e19a563cb7074ed7c69d33
+size 507210373