Upload app.py

app.py

@@ -7,10 +7,29 @@ import gradio as gr
 import numpy as np
 import pandas as pd
 import matplotlib.pyplot as plt
-from transformers import BertConfig, BertTokenizer, XLMRobertaForSequenceClassification
+from transformers import BertConfig, BertTokenizer, XLMRobertaForSequenceClassification, BertForTokenClassification
 from keras.models import load_model
 
 
+def text_clf_ori(text):
+    vocab_file = "vocab.txt"  # 词汇表
+    tokenizer = BertTokenizer(vocab_file)
+    # 加载模型
+    config = BertConfig.from_pretrained("nanaaaa/emotion_chinese_english")
+    model = BertForTokenClassification.from_pretrained("nanaaaa/emotion_chinese_english", config=config)
+
+    inputs = tokenizer(text, return_tensors="pt")
+
+    # 模型推断
+    outputs = model(**inputs)
+    probs = torch.nn.functional.softmax(outputs.logits, dim=1)
+    # 创建标签和概率列表
+    labels = ["害怕", "高兴喵", "惊喜", "伤心", "生气"]
+    probabilities = probs.detach().cpu().numpy()[0].tolist()
+    # 返回标签和概率列表
+    return {labels[i]: float(probabilities[0][i]) for i in range(len(labels))}
+
+
 def text_clf(text):
     vocab_file = "vocab.txt"  # 词汇表
     tokenizer = BertTokenizer(vocab_file)
@@ -125,7 +144,9 @@ with gr.Blocks() as demo:
     with gr.Tab("Flip Text"):
         text = gr.Textbox(label="文本哟")
         text_output = gr.outputs.Label(label="情感呢")
+        text_output1 = gr.outputs.Label(label="情感呢")
         text_button = gr.Button("确认")
+        text_button1 = gr.Button("确认对比")
 
     with gr.Tab("Flip Audio"):
         audio = gr.Audio(label="音频捏")
@@ -139,6 +160,7 @@ with gr.Blocks() as demo:
         cir_button = gr.Button("确认")
 
     text_button.click(fn=text_clf, inputs=text, outputs=text_output)
+    text_button1.click(fn=text_clf_ori, inputs=text, outputs=text_output1)
     audio_button.click(fn=audio_clf, inputs=audio, outputs=audio_output)
     cir_button.click(fn=cir_clf, inputs=[cir_l, cir_r], outputs=cir_output)