Upload 3 files

application.py

@@ -0,0 +1,91 @@
+import numpy as np
+import torch
+import transformers
+import json
+from flask import Flask, jsonify, request
+import torch.nn.functional as F
+import boto3
+import pandas as pd
+bucket = 'data-ai-dev2'
+from transformers import BertTokenizer, BertModel
+from torch import cuda
+device = 'cuda' if cuda.is_available() else 'cpu'
+
+class RobertaClass(torch.nn.Module):
+    def __init__(self):
+        super(RobertaClass, self).__init__()
+        self.l1 = BertModel.from_pretrained("bert-base-multilingual-cased")
+        self.pre_classifier = torch.nn.Linear(768, 768)
+        self.dropout = torch.nn.Dropout(0.3)
+        self.classifier = torch.nn.Linear(768, 8)
+
+    def forward(self, input_ids, attention_mask, token_type_ids):
+        output_1 = self.l1(input_ids=input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids)
+        hidden_state = output_1[0]
+        pooler = hidden_state[:, 0]
+        pooler = self.pre_classifier(pooler)
+        pooler = torch.nn.ReLU()(pooler)
+        pooler = self.dropout(pooler)
+        output = self.classifier(pooler)
+        return output
+
+model = RobertaClass()
+model.to(device)    
+
+s3 = boto3.client('s3', aws_access_key_id='AKIAW5BGUY6ZRCSQBSIJ', aws_secret_access_key= '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')
+s3.download_file(Bucket=bucket, Key='model_hf/tweet_model/tweet_model_v1.bin', Filename = './tweet_model_v1.bin')
+
+model = torch.load('tweet_model_v1.bin', map_location=torch.device('cpu'))
+
+tokenizer = BertTokenizer.from_pretrained('bert-base-multilingual-cased', truncation=True, do_lower_case=True)
+
+def id2class_fun(lst, map_cl):
+    s = pd.Series(lst)
+    return s.map(map_cl).tolist()
+    
+application = Flask(__name__)
+@application.route('/')
+def home():
+    return "Working!"
+
+@application.route('/process/', methods=['POST'])
+def process():
+    content_type = request.headers.get('Content-Type')
+    if (content_type == 'application/json'):
+        json_file = request.json
+    loaded = json.dumps(json_file)
+    new_loaded = json.loads(loaded)
+    text = new_loaded['text']
+    id2class = {0: 'InappropriateUndesirable', 1 : 'GreenContent', 2 : 'IllegalActivities', 
+                3 : 'DiscriminatoryHate', 4 :'ViolentGraphic', 5:'PotentialAddiction', 
+                6 : 'ExtremismTerrorism', 7 : 'SexualExplicit'}
+    try:
+
+        inputs = (
+            tokenizer.encode_plus(
+                text, None, add_special_tokens=True, max_length = 512,
+                return_token_type_ids=True, padding=True, 
+                truncation=True, return_tensors='pt'))
+        ids = inputs['input_ids']
+        mask = inputs['attention_mask']
+        token_type_ids = inputs["token_type_ids"]
+        outputs = model(ids, mask, token_type_ids)
+        top_values, top_indices = torch.topk(outputs.data, k=2, dim=1)
+        probs_values = F.softmax(top_values, dim=0)
+        prd_cls = top_indices.cpu().detach().numpy().tolist()
+        prd_cls = [item for sublist in prd_cls for item in sublist]
+        prd_cls_1 = id2class_fun(prd_cls, id2class)
+        prd_score = top_values.cpu().detach().numpy().tolist()
+        prd_score = [item for sublist in prd_score for item in sublist]
+        otp = dict(zip(prd_cls_1, prd_score))
+    #     .replace(map_class, inplace=True)
+        return jsonify({'output':otp})
+    except:
+        return jsonify({'output':'something went wrong'})
+
+     
+        
+if __name__ == "__main__":
+    application.debug = True
+    application.run()
+

requirements.txt

@@ -0,0 +1,6 @@
+transformers==4.31.0
+numpy==1.25.2
+Flask==2.3.2
+boto3==1.26.157
+torch==2.0.0
+pandas==1.5.3

tweet_model_v1.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaa1adad810a4ec32ba1e5e7226eafc7f083953355d902d5d67cfebab2a72359
+size 713927888