upload api codes

Dockerfile

@@ -0,0 +1,13 @@
+FROM python:3.9
+
+RUN useradd -m -u 1000 user
+USER user
+ENV PATH="/home/user/.local/bin:$PATH"
+
+WORKDIR /app
+
+COPY --chown=user ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+COPY --chown=user . /app
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,113 @@
+import torch
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+import torch.nn as nn
+import pickle
+import re
+
+token_2_id = None
+# Load the dictionary later
+with open(r"vocab.pkl", "rb") as f:
+    token_2_id = pickle.load(f)
+print(token_2_id)
+
+
+def normalize(text):
+    text = text.lower()
+    text = re.sub(r'[^a-z0-9\s]', '', text)
+    text = ' '.join(text.split())
+    return text
+
+
+def tokenize(text):
+    tokens = text.split()
+    return tokens
+
+
+def convert_tokens_2_ids(tokens):
+    input_ids = [
+        token_2_id.get(token, token_2_id['<UNK>']) for token in tokens
+    ]
+    return input_ids
+
+
+def process_text(text, aspect):
+    text_aspect_pair = text + ' ' + aspect
+    normalized_text = normalize(text_aspect_pair)
+    tokens = tokenize(normalized_text)
+    input_ids = convert_tokens_2_ids(tokens)
+    input_ids = torch.tensor(input_ids).unsqueeze(0)
+    return input_ids
+
+
+class ABSA(nn.Module):
+    def __init__(self, vocab_size, num_labels=3):
+        super(ABSA, self).__init__()
+        self.vocab_size = vocab_size
+        self.num_labels = num_labels
+        self.embedding_layer = nn.Embedding(
+            num_embeddings=vocab_size, embedding_dim=256
+        )
+        self.lstm_layer = nn.LSTM(
+            input_size=256,
+            hidden_size=512,
+            batch_first=True,
+        )
+
+        self.fc_layer = nn.Linear(
+            in_features=512,
+            out_features=self.num_labels
+        )
+
+    def forward(self, x):
+        embeddings = self.embedding_layer(x)
+        lstm_out, _ = self.lstm_layer(embeddings)
+        logits = self.fc_layer(lstm_out[:, -1, :])
+        return logits
+
+
+model = ABSA(vocab_size=len(token_2_id.keys()), num_labels=3)
+model.load_state_dict(torch.load(r'model_weights.pth'))
+model.eval()
+
+print("Model loaded successfully")
+
+app = FastAPI()
+
+
+# space url = "https://masumbhuiyan-myabsaservices.hf.space/"
+# greet api = "https://masumbhuiyan-myabsaservices.hf.space/greet"
+# post api = "https://masumbhuiyan-myabsaservices.hf.space/predict"
+@app.get("/greet")
+def greet_json():
+    return {"message": "Hello World"}
+
+
+class TextAspectInput(BaseModel):
+    text: str
+    aspect: str
+
+
+SENTIMENT_LABELS = {0: "Negative", 1: "Neutral", 2: "Positive"}
+
+
+@app.post("/predict")
+async def predict_sentiment(input_data: TextAspectInput):
+    try:
+        text = input_data.text
+        aspect = input_data.aspect
+        input_ids = process_text(text, aspect)
+
+        try:
+            with torch.no_grad():
+                logits = model(input_ids)
+                probs = torch.softmax(logits, dim=-1)
+                prediction = probs.argmax(dim=-1).item()
+                sentiment = SENTIMENT_LABELS[prediction]
+        except Exception as e:
+            raise Exception(e)
+
+        return {"sentiment": sentiment, "probabilities": probs.squeeze().tolist()}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+

model_weights.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b1a6752d09f7a562dd93c6d857987c4df789908e132287bef226ac82c6c2a80
+size 10208198

requirements.txt

@@ -0,0 +1,5 @@
+fastapi
+uvicorn[standard]
+numpy
+Pillow
+torch

vocab.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e195b47d3a84c730ff3a7f25df52b335174ba06914d9404bffa7f4422603d60
+size 47213