Update proton_code_llama_fine_tuning_using_qlora_inferencing.py

proton_code_llama_fine_tuning_using_qlora_inferencing.py

@@ -7,13 +7,13 @@ Original file is located at
     https://colab.research.google.com/drive/1snQ0jgdbZDXNuVpvryjfENfCjk8K2SE2
 """
 
-!pip install -q -U trl transformers accelerate git+https://github.com/huggingface/peft.git
-!pip install -q datasets bitsandbytes einops wandb
-!pip install peft
-!pip install --upgrade gradio
-!pip install sentence_transformers
-!pip install accelerate
-!pip install bitsandbytes
+# !pip install -q -U trl transformers accelerate git+https://github.com/huggingface/peft.git
+# !pip install -q datasets bitsandbytes einops wandb
+# !pip install peft
+# !pip install --upgrade gradio
+# !pip install sentence_transformers
+# !pip install accelerate
+# !pip install bitsandbytes
 
 import pandas as pd
 import torch
@@ -21,11 +21,14 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
 from peft import PeftModel, PeftConfig
 from sentence_transformers.util import semantic_search
 import requests
+from datasets import load_dataset
 
 model_id = "sentence-transformers/all-MiniLM-L6-v2"
 hf_token = "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"
 
-data = pd.read_excel('/content/sample_data/Proton Gen AI.xlsx')
+# data = pd.read_excel('/content/sample_data/Proton Gen AI.xlsx')
+
+dataset = load_dataset("Siva1995/Proton_GENAI")
 
 text_col = []
 
@@ -99,45 +102,45 @@ def getExamples(question):
     return r
 
 def prepare_prompt(query):
-  prompt_prep = ''.join([priming_text,getExamples(query),'\n',decorator,'\n','Similar to the above examples, generate only one Proton UI for the following description:',query])
-  return prompt_prep
+    prompt_prep = ''.join([priming_text,getExamples(query),'\n',decorator,'\n','Similar to the above examples, generate only one Proton UI for the following description:',query])
+    return prompt_prep
 
 def generate(query):
-  prompt_prep = prepare_prompt(query)
-  encodings = model_tokenizer(prompt_prep, return_tensors = 'pt').to(device)
-
-  with torch.inference_mode():
-    outputs = model.generate(input_ids=encodings.input_ids,
+    prompt_prep = prepare_prompt(query)
+    encodings = model_tokenizer(prompt_prep, return_tensors = 'pt').to(device)
+    
+    with torch.inference_mode():
+        outputs = model.generate(input_ids=encodings.input_ids,
                             attention_mask = encodings.attention_mask,
                             generation_config = generation_config,
                           max_new_tokens=1000,do_sample=True)
-
-  outputs = model_tokenizer.decode(outputs[0],skip_special_tokens=True,best_of=1)
-
-  return outputs[outputs.find(query): ]#outputs.find('Explanation')
-
-output = generate('Create a choice type of Proton UI for choosing vegetables')
-print(output)
-
-import gradio as gr
-with gr.Blocks() as demo:
-    gr.Markdown("****This application is built on fine-tuned on specific data on codellama 7b on 4 bit fine tuned****")
-    gr.Markdown("**No Paid service has been leveraged**")
-    gr.Markdown("**Type your Query Run to see the output**")
-    with gr.Row():
-        out = gr.Textbox()
-    with gr.Row():
-        inp = gr.Textbox(placeholder="Type your Question")
-    btn = gr.Button("Run")
-    btn.click(fn=output, inputs=inp, outputs=out)
-demo.launch(auth=("user", "password"),share=True)
-
-import random
-import gradio as gr
-
-def alternatingly_agree(message, history):
-  output = generate(str(message))
-  return output
-
-gr.ChatInterface(alternatingly_agree).launch()
+    
+    outputs = model_tokenizer.decode(outputs[0],skip_special_tokens=True,best_of=1)
+    
+    return outputs[outputs.find(query): ]#outputs.find('Explanation')
+
+# output = generate('Create a choice type of Proton UI for choosing vegetables')
+# print(output)
+
+# import gradio as gr
+# with gr.Blocks() as demo:
+#     gr.Markdown("****This application is built on fine-tuned on specific data on codellama 7b on 4 bit fine tuned****")
+#     gr.Markdown("**No Paid service has been leveraged**")
+#     gr.Markdown("**Type your Query Run to see the output**")
+#     with gr.Row():
+#         out = gr.Textbox()
+#     with gr.Row():
+#         inp = gr.Textbox(placeholder="Type your Question")
+#     btn = gr.Button("Run")
+#     btn.click(fn=output, inputs=inp, outputs=out)
+# demo.launch(auth=("user", "password"),share=True)
+
+# import random
+# import gradio as gr
+
+# def alternatingly_agree(message, history):
+#   output = generate(str(message))
+#   return output
+
+# gr.ChatInterface(alternatingly_agree).launch()