Update app.py

app.py

@@ -7,24 +7,26 @@ from gradio.themes.utils import colors, fonts, sizes
 from llama_cpp import Llama
 from huggingface_hub import hf_hub_download
 
-hf_hub_download(repo_id="TheBloke/dolphin-2.5-mixtral-8x7b-GGUF", filename="dolphin-2.5-mixtral-8x7b.Q2_K.gguf", local_dir=".")
-llm = Llama(model_path="./dolphin-2.5-mixtral-8x7b.Q2_K.gguf")
+hf_hub_download(repo_id="cognitivecomputations/dolphin-2.9-llama3-8b-gguf", filename="dolphin-2.9-llama3-8b-q3_K_M.gguf", local_dir=".")
+llm = Llama(model_path="./dolphin-2.9-llama3-8b-q3_K_M.gguf")
 
-ins = '''<|im_start|>user
+ins = '''<|im_start|>system
+{system}<|im_end|>
+<|im_start|>user
 {question}<|im_end|>
 <|im_start|>assistant
 '''
 
 theme = gr.themes.Monochrome(
-	primary_hue="blue",
-	secondary_hue="cyan",
+	primary_hue="red",
+	secondary_hue="orange",
 	neutral_hue="neutral",
 	radius_size=gr.themes.sizes.radius_sm,
 	font=[gr.themes.GoogleFont("Space Grotesk"), "ui-sans-serif", "system-ui", "sans-serif"],
 )
 
-def generate(instruction):
-	prompt = ins.format(question=instruction)
+def generate(instruction, system_prompt):
+	prompt = ins.format(question=instruction, system=system_prompt)
 	response = llm(prompt, stop=['<|im_start|>user', '<|im_end|>'])
 	result = response['choices'][0]['text']
 	return result
@@ -95,16 +97,17 @@ custom_theme = BlueTheme()
 with gr.Blocks(theme=custom_theme, analytics_enabled=False, css=css) as demo:
 	with gr.Column():
 		gr.Markdown(
-			""" # 🐬 Dolphin4ALL
+			""" # 🦙 Dolphin4ALL
 
-			8x7b quantized 2bit (q2_k)
+			llama3 8b (q3_k_m)
 
 			Type in the box below and click the button to generate answers to your most pressing questions!
 		""")
 
 		with gr.Row():
 			with gr.Column(scale=3):
-				instruction = gr.Textbox(placeholder="Enter your question here", label="Question", elem_id="q-input")
+				instruction = gr.Textbox(placeholder="Enter your question here", label="Question Prompts")
+				sys_prompt = gr.Textbox(placeholder="Enter your system instructions here", label="System Prompts")
 
 				with gr.Box():
 					gr.Markdown("**Answer**")
@@ -118,7 +121,7 @@ with gr.Blocks(theme=custom_theme, analytics_enabled=False, css=css) as demo:
 					outputs=[output],
 				)
 
-	submit.click(generate, inputs=[instruction], outputs=[output])
-	instruction.submit(generate, inputs=[instruction], outputs=[output])
+	submit.click(generate, inputs=[instruction, sys_prompt], outputs=[output])
+	instruction.submit(generate, inputs=[instruction, sys_prompt], outputs=[output])
 
 demo.queue(concurrency_count=1).launch(debug=True)