Create app.py

app.py

@@ -0,0 +1,62 @@
+import gradio as gr
+from datasets import load_dataset, load_from_disk, Dataset
+from transformers import AutoTokenizer, AutoModel
+import torch
+import pandas as pd
+import base64
+
+styleBase64 = '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'
+style = base64.b64decode(styleBase64).decode("ascii")
+
+model_ckpt = "nomic-ai/nomic-embed-text-v1.5"
+
+tokenizer = AutoTokenizer.from_pretrained(model_ckpt)
+model = AutoModel.from_pretrained(model_ckpt, trust_remote_code=True)
+
+device = torch.device("cpu")
+model.to(device)
+
+
+def cls_pooling(model_output):
+    return model_output.last_hidden_state[:, 0]
+
+def get_embeddings(text_list):
+    encoded_input = tokenizer(
+        text_list, padding=True, truncation=True, return_tensors="pt"
+    )
+    encoded_input = {k: v.to(device) for k, v in encoded_input.items()}
+    model_output = model(**encoded_input)
+    return cls_pooling(model_output)
+
+
+embeddings_dataset = load_dataset("Vadim212/doctest1")["train"]
+embeddings_dataset.add_faiss_index(column="embeddings")
+
+def find(question):
+
+    question_embedding = get_embeddings([question]).cpu().detach().numpy()
+
+    scores, samples = embeddings_dataset.get_nearest_examples(
+        "embeddings", question_embedding, k=5
+    )
+
+    samples_df = pd.DataFrame.from_dict(samples)
+    samples_df["scores"] = scores
+    samples_df.sort_values("scores", ascending=True, inplace=True)
+    
+    result = [(style + f"<font size='20'><b>Answer {i}:</b></font>" + row.url + "<hr>") for i, row in samples_df.iterrows()]
+    return result
+
+
+demo = gr.Blocks()
+
+with demo:
+    inp = gr.Textbox(placeholder="Enter prompt", label= "Prompt like: 'how to export to PDF?', 'What is Stimulsoft Reports?', 'What is Dashboards?', 'how to render report?', 'List of exports'" )
+    find_btn = gr.Button("Find")
+    big_block = [gr.HTML("") for i in range(5)]
+    
+    find_btn.click(find, 
+               inputs=inp, 
+               outputs=big_block)
+
+demo.launch()