first look

.gitignore

@@ -0,0 +1,2 @@
+sb-rag/
+*.pyc

.vscode/launch.json

@@ -0,0 +1,23 @@
+{
+    // Use IntelliSense to learn about possible attributes.
+    // Hover to view descriptions of existing attributes.
+    // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+    "version": "0.2.0",
+    "configurations": [
+
+        {
+            "name": "Python Debugger: Current File with Arguments",
+            "type": "debugpy",
+            "request": "launch",
+            "cwd": "${workspaceFolder}",
+            "program": "${workspaceFolder}/sb-rag/Scripts/gradio.exe",
+            "console": "integratedTerminal",
+            "args": [
+                "app.py"
+            ],
+            "env": {
+                "HF_SERVELESS_API": "WRITE THIS"
+            }
+        }
+    ]
+}

README.md

@@ -11,3 +11,22 @@ short_description: Just another rag but with Images 🖼️
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+
+## Local Debug
+
+Highly recommend VS Code, makes life easy.
+
+1. Create Virtual environment with name `sb-rag` using the below command
+    `python -m venv sb-rag`
+
+2. Activate the environemnt (create new terminal VS Code to automatically do so)
+
+3. Edit `.vscode/launch.json`. Fill in the environment variable `HF_SERVERLESS_API`.
+
+4. Start VS Code debugger.
+
+## References
+
+1. UI Blocks Concepts: https://huggingface.co/learn/nlp-course/en/chapter9/7
+2. UI Row-Column Arrangement: https://www.gradio.app/guides/controlling-layout
+3. Show caption in image gallery: https://github.com/gradio-app/gradio/issues/3364

app.py

@@ -0,0 +1,71 @@
+import numpy as np
+import gradio as gr
+from z_generate import ServerlessInference
+
+# STATIC TEXT DISPLAY
+TXT_APP_DESCRIPTION = '''
+Just another Retrieval Augmented Generation that also retrieves images
+'''
+TXT_SOURCE_DOC_DESCRIPTION = '''
+Manually Downloaded as HTML files: 
+
+1. https://en.wikipedia.org/wiki/MS_Dhoni
+2. https://en.wikipedia.org/wiki/Jharkhand
+2. https://en.wikipedia.org/wiki/Cricket_World_Cup
+'''
+
+
+# UI Interface
+demo = gr.Blocks()
+
+llm = ServerlessInference()
+
+# Processing Functions
+def update_response(query:str = "something"):
+    return llm.test(query)
+    
+def update_gallery(text:str = "hell"):
+    imgs = [
+                ("http://www.marketingtool.online/en/face-generator/img/faces/avatar-1151ce9f4b2043de0d2e3b7826127998.jpg", "Some Description"),
+                ("http://www.marketingtool.online/en/face-generator/img/faces/avatar-116b5e92936b766b7fdfc242649337f7.jpg", "Another Description")
+            ]
+    return imgs
+
+
+def ask_bot(text):
+    return update_response(text), update_gallery(text)
+
+# UI Layout
+with demo:
+    gr.Markdown(TXT_APP_DESCRIPTION)
+
+    with gr.Tabs():
+        with gr.TabItem("Ask Bot"):
+            with gr.Row(equal_height=True):
+                with gr.Column(scale=3):
+                    text_input = gr.Textbox(
+                        label="You query here",
+                        placeholder="What positions apart from crickter did Dhoni held?"
+                    )
+                with gr.Column(scale=1):
+                    btn_generate = gr.Button("Generate Answer")
+
+            with gr.Row():
+                with gr.Column(scale=3):
+                    text_output = gr.Textbox(label="Bot Response:", placeholder="Type in Query before I could answer")
+                
+                with gr.Column(scale=2): 
+                    gallery = gr.Gallery(
+                        label="Generated images", show_label=False, elem_id="gallery"
+                        , columns=[3], rows=[1], object_fit="contain", height="auto"
+                    )
+
+            btn_generate.click(ask_bot, text_input, outputs=[text_output, gallery])
+
+
+
+        ####
+        with gr.TabItem("Source Documents"):
+            gr.Markdown(TXT_SOURCE_DOC_DESCRIPTION)
+
+demo.launch(debug=True)

z_generate.py

@@ -0,0 +1,23 @@
+from huggingface_hub import InferenceClient
+import os 
+
+class ServerlessInference:
+    def __init__(self):
+        self.model:str = "HuggingFaceH4/zephyr-7b-beta"
+        self.client = InferenceClient(api_key=os.getenv("HF_SERVELESS_API"))
+
+    def test(self, query:str) -> str:
+        '''Responds to query using llm'''
+        messages:str = [
+            {
+                "role": "user",
+                "content": query
+            }
+        ]
+        completion = self.client.chat.completions.create(
+            model=self.model,
+            messages=messages, 
+            max_tokens=500
+        )
+
+        return completion.choices[0].message.content