simple implementation of flags

README.md

@@ -32,7 +32,8 @@ source dschatbot/bin/activate
 2. Run the chatbot:
 
     ```bash
-    python app.py
+    source .env
+    sdkKEY=${sdkKEY} python app.py
     ```
 
 ## Usage

app.py

@@ -1,3 +1,4 @@
+from ldclient import Context
 import os
 import logging
 import gradio as gr
@@ -5,6 +6,12 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 import uuid
 import time
+import ldclient
+
+from ldclient.config import Config
+
+ldclient.set_config(Config(os.environ.get('sdkKEY')))
+client = ldclient.get()
 
 
 def capture_logs(log_body, log_file, uuid_label):
@@ -35,6 +42,24 @@ print("CUDA available: ", torch.cuda.is_available())
 print("MPS available: ", torch.backends.mps.is_available())
 
 
+context = Context.builder("huggie-face") \
+    .set("application", "deepSeekChatbot") \
+    .build()
+
+flag_value = client.variation("themeColors", context, False)
+if flag_value:
+    print("Feature flag on")
+    theme = gr.themes.Soft(
+        primary_hue="fuchsia",
+        neutral_hue="blue",
+    )
+else:
+    print("Feature flag off")
+    theme = gr.themes.Soft(
+        primary_hue="sky",
+        neutral_hue="slate",
+    )
+
 tokenizer = AutoTokenizer.from_pretrained(
     "deepseek-ai/deepseek-coder-1.3b-instruct", trust_remote_code=True)
 model = AutoModelForCausalLM.from_pretrained(
@@ -49,17 +74,8 @@ device = torch.device(
     "mps") if torch.backends.mps.is_available() else torch.device("cpu")
 model = model.to(device)
 
-# Theme builder
-# gr.themes.builder()
-
-theme = gr.themes.Soft(
-    primary_hue="sky",
-    neutral_hue="slate",
-)
 
 # Function to handle user input and generate a response
-
-
 def chatbot_response(query, tokens, top_k, top_p):
     uuid_label = str(uuid.uuid4())
 
@@ -83,9 +99,10 @@ def chatbot_response(query, tokens, top_k, top_p):
         outputs[0][len(inputs[0]):], skip_special_tokens=True)
 
     end_time = time.time()  # End timer
-    performance_time = round(end_time - start_time, 2)    
+    performance_time = round(end_time - start_time, 2)
 
-    log_body = 'query: %s, pocessTime: %s,  tokens: %s, top_k: %s, top_p: %s' % (query, performance_time, tokens, top_k, top_p)
+    log_body = 'query: %s, pocessTime: %s,  tokens: %s, top_k: %s, top_p: %s' % (
+        query, performance_time, tokens, top_k, top_p)
 
     capture_logs(uuid_label, 'query_logs.csv', log_body)
 

requirements.txt

@@ -5,4 +5,5 @@ transformers
 minijinja
 torch --extra-index-url https://download.pytorch.org/whl/cu118
 torchvision
-torchaudio
+torchaudio
+launchdarkly-server-sdk