Create app.py

app.py

@@ -0,0 +1,62 @@
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, pipeline
+
+import time
+import gradio as gr
+
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_compute_dtype=torch.bfloat16,
+)
+
+model = AutoModelForCausalLM.from_pretrained(
+    "microsoft/Phi-3.5-mini-instruct",
+    torch_dtype=torch.bfloat16,
+    quantization_config=bnb_config,
+    trust_remote_code=True
+)
+model.load_adapter('./finetunedPEFTModel')
+
+tokenizer = AutoTokenizer.from_pretrained('./finetunedPEFTModel', trust_remote_code=True)
+# tokenizer.pad_token = tokenizer.unk_token
+# tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-3.5-mini-instruct", trust_remote_code=True)
+
+
+def generateText(inputText="JULIET\n", num_tokens=200):
+
+    # pipe = pipeline(task="text-generation", model=model, tokenizer=tokenizer, max_length=num_tokens)
+    # result = pipe(f'''[INST] {inputText} [/INST]''')
+    # print(result[0]['generated_text'])
+    
+    prompt = "What is model regularization?"
+    pipe = pipeline(task="text-generation", model=model, tokenizer=tokenizer, max_length=num_tokens)
+    result = pipe(f'''{inputText}''')
+    
+    return result[0]['generated_text']
+    
+
+title = "Fine tuned Phi3.5 instruct model on OpenAssist dataset using QLora"
+description = "Fine tuned Phi3.5 instruct model on OpenAssist dataset using QLora"
+examples = [
+            ["How can I optimize my web page for online search so that it is on top?", 200],
+            ["Can you give me an example of python script for Fibonacci series?", 200],
+            ["Can you explain what is Contrastive Loss in Deep Learning?", 200],
+            ["How are Sentence Transformers different from Huggingface Transformers?", 200],
+           ]
+
+demo = gr.Interface(
+    generateText, 
+    inputs = [
+        gr.Textbox(label="Question that you want to ask"),
+        gr.Slider(100, 500, value = 200, step=100, label="Number of chars that you want in your output"),
+        ], 
+    outputs = [
+        gr.Text(),
+        ],
+    title = title,
+    description = description,
+    examples = examples,
+    cache_examples=False    
+)
+demo.launch()