Upload app.py

app.py

@@ -1,4 +1,80 @@
 import streamlit as st
+from torch.utils.data import Dataset, DataLoader
+import torch
+from sklearn.model_selection import train_test_split
+from transformers import get_linear_schedule_with_warmup, AdamW
+from torch.cuda.amp import autocast, GradScaler
+from transformers import DistilBertForSequenceClassification, DistilBertTokenizer, \
+                         BigBirdPegasusForSequenceClassification, BigBirdTokenizer
+from transformers import pipeline
+from torch.utils.data import TensorDataset, random_split, DataLoader, RandomSampler, SequentialSampler
+from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score
+import streamlit as st
+
+import pandas as pd
+import json
+import ast
+from scipy import stats
+import numpy as np
+import time
+import datetime
+
+#
+def get_top95(y_predict, convert_target):
+    lst_labels = []
+    tuple_arr = tuple((idx, val) for idx, val in enumerate(y_predict[0]))
+    sort_y = sorted(tuple_arr, key=lambda x: x[1], reverse=True)
+    cumsum = 0
+    for key, prob in sort_y:
+        cumsum += prob
+        print(prob)
+        lst_labels.append(convert_target[str(key)])
+        if cumsum > 0.95:
+            break
+    return lst_labels
+#
+# model = MyModel()
+model = torch.load("distilbert-model1.pt", map_location='cpu').eval()
+# print(model)
+# model = DistilBertForSequenceClassification.from_pretrained("model/distilbert-model1.pt", local_files_only=True)
+# tokenizer = BigBirdTokenizer.from_pretrained('google/bigbird-pegasus-large-arxiv')
+
+# model = BigBirdPegasusForSequenceClassification.from_pretrained('google/bigbird-pegasus-large-arxiv',
+#     num_labels=8,
+#     return_dict=False)
+
+def get_predict(text):
+    tokenizer = DistilBertTokenizer.from_pretrained('distilbert-base-cased')
+    # encoded_dict = tokenizer.encode_plus(
+    #     text,  # document to encode.
+    #     add_special_tokens=True,  # add '[CLS]' and '[SEP]'
+    #     max_length=512,  # set max length
+    #     truncation=True,  # truncate longer messages
+    #     pad_to_max_length=True,  # add padding
+    #     return_attention_mask=True,  # create attn. masks
+    #     return_tensors='pt'  # return pytorch tensors
+    # )
+
+    inputs = tokenizer(text, return_tensors="pt")
+    outputs = model(
+        input_ids=inputs['input_ids'],
+        attention_mask=inputs['attention_mask'],
+    )
+    logits = outputs[0]
+    y_predict = torch.nn.functional.softmax(logits).cpu().detach().numpy()
+    file_path = "decode_target (1).json"
+
+    with open(file_path, 'r') as json_file:
+        decode_target = json.load(json_file)
+    print(get_top95(y_predict, decode_target))
+#
+#
+#
+#
+#
+# get_predict('''physics physics physics physics physics
+#                physics physics physics physics''')
+#
 
 st.markdown("### Hello, world!")
 st.markdown("<img width=200px src='https://rozetked.me/images/uploads/dwoilp3BVjlE.jpg'>", unsafe_allow_html=True)
@@ -12,4 +88,4 @@ text = st.text_area("TEXT HERE")
 # raw_predictions = pipe(text)
 # тут уже знакомый вам код с huggingface.transformers -- его можно заменить на что угодно от fairseq до catboost
 
-st.markdown("It's prediction: GEs")
+st.markdown("It's prediction: {get_predict(text)}")