deploy code

app.py

@@ -0,0 +1,36 @@
+import gradio as gr
+from transformers import AutoTokenizer
+import torch
+from model import EnergySmellsDetector
+from config import SMELLS, BEST_THRESHOLD
+
+TOKENIZER = "microsoft/graphcodebert-base"
+tokenizer = AutoTokenizer.from_pretrained(TOKENIZER)
+model = EnergySmellsDetector.load_model_from_hf()
+
+
+def round_logit(logits, threshold):
+    logits = (logits > threshold).to(int)
+    return logits.cpu().numpy()
+
+
+def greet(code_snippet):
+    inputs = tokenizer(code_snippet, return_tensors="pt", truncation=True)
+    with torch.no_grad():
+        logits = model(**inputs)[0]
+        rounded_logits = round_logit(logits, BEST_THRESHOLD)
+        return f"{dict(zip(SMELLS, map(int, rounded_logits)))}"
+
+
+
+textbox = gr.Textbox(label="Enter your code snippet", placeholder="Here goes your code")
+description = "An application to identify whether your code has energy smells or not. It predicts the presence of 9 different energy smells."
+title = "Energy Smells Detector"
+
+gr.Interface(
+    title=title,
+    description=description,
+    inputs=textbox,
+    fn=greet,
+    outputs="text"
+).launch()

config.py

@@ -0,0 +1,32 @@
+import torch
+from transformers import set_seed
+from pathlib import Path
+import os
+
+
+    
+
+SMELLS = [
+    "BuiltInBehavior",
+    "ExtraUsageOfArrays",
+    "ImproperDataStructure",
+    "LackingBreak",
+    "NotCaching",
+    "OveruseOfLoops",
+    "UnnecessaryConditionals",
+    "UnnecessaryRecursion",
+    "UnnecessaryTypeConversion",
+]
+
+
+# Configurations
+DROPOUT_PROB = 0.1
+HIDDEN_SIZE = 768
+NUM_LABELS = 9
+LAST_NUM_NEURON = 9
+BEST_THRESHOLD = 0.3
+PRETRAINED_MODEL = "microsoft/graphcodebert-base"
+MODEL_SAVE_DIR = Path(__file__).parents[1] / "models_weights"
+WEIGHTS_FILE_NAME = "weights.bin"
+MODEL_SAVE_PATH = MODEL_SAVE_DIR / WEIGHTS_FILE_NAME
+HF_REPO_NAME = f"mjavadmt/EnergySmell"

model.py

@@ -0,0 +1,36 @@
+import torch
+import torch.nn as nn
+from transformers import AutoModel
+from config import (
+    HIDDEN_SIZE,
+    DROPOUT_PROB,
+    LAST_NUM_NEURON,
+    HF_REPO_NAME,
+    WEIGHTS_FILE_NAME,
+    PRETRAINED_MODEL,
+)
+from huggingface_hub import hf_hub_download
+
+
+class EnergySmellsDetector(nn.Module):
+    def __init__(self, model_name):
+        super(EnergySmellsDetector, self).__init__()
+        self.model = AutoModel.from_pretrained(model_name)
+        self.dropout = nn.Dropout(DROPOUT_PROB)
+        self.fc = nn.Linear(HIDDEN_SIZE, LAST_NUM_NEURON)
+
+    def forward(self, input_ids, attention_mask):
+        outputs = self.model(input_ids=input_ids, attention_mask=attention_mask)
+        x = self.dropout(outputs.pooler_output)
+        logits = self.fc(x)
+        return torch.sigmoid(logits).to(float)
+
+    @staticmethod
+    def load_model_from_hf():
+        model_path = hf_hub_download(repo_id=HF_REPO_NAME, filename=WEIGHTS_FILE_NAME)
+
+        # Load model
+        model = EnergySmellsDetector(PRETRAINED_MODEL)
+        model.load_state_dict(torch.load(model_path))
+
+        return model

requirements.txt

@@ -0,0 +1,2 @@
+transformers
+torch