Create train.py

train.py

@@ -0,0 +1,64 @@
+import pandas as pd
+import torch
+import re
+from datasets import Dataset
+from transformers import (
+    AutoModelForTokenClassification,
+    AutoTokenizer,
+    Trainer,
+    TrainingArguments,
+    DataCollatorForTokenClassification,
+)
+from huggingface_hub import notebook_login
+
+# Login to Hugging Face Hub (Make sure your Space is set to private if needed)
+notebook_login()
+
+# Step 1: Load Luxury Fashion Dataset (Replace with actual dataset)
+df = pd.read_csv("luxury_apparel_data.csv")  # Update with correct dataset file
+
+# Keep only relevant columns
+df = df[['brand', 'category', 'description', 'price']].dropna()
+
+# Generate search queries from dataset
+df['query'] = df.apply(lambda x: f"{x['brand']} {x['category']} under {x['price']} AED", axis=1)
+
+# Step 2: Tokenization
+model_name = "dslim/bert-base-NER"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+
+def tokenize_batch(batch):
+    return tokenizer(batch['query'], padding=True, truncation=True)
+
+# Convert dataframe into Hugging Face dataset
+hf_dataset = Dataset.from_pandas(df[['query']])
+hf_dataset = hf_dataset.map(tokenize_batch, batched=True)
+
+# Step 3: Fine-tune the Pretrained NER Model
+model = AutoModelForTokenClassification.from_pretrained(model_name)
+
+training_args = TrainingArguments(
+    output_dir="./luxury_ner_model",
+    evaluation_strategy="epoch",
+    save_strategy="epoch",
+    per_device_train_batch_size=8,
+    per_device_eval_batch_size=8,
+    num_train_epochs=3,
+    logging_dir="./logs",
+    logging_steps=500,
+)
+
+trainer = Trainer(
+    model=model,
+    args=training_args,
+    train_dataset=hf_dataset,
+    eval_dataset=hf_dataset,
+    tokenizer=tokenizer,
+    data_collator=DataCollatorForTokenClassification(tokenizer),
+)
+
+trainer.train()
+
+# Save model to Hugging Face Hub
+model.push_to_hub("luxury-fashion-ner")
+tokenizer.push_to_hub("luxury-fashion-ner")