diff --git a/.gitignore b/.gitignore
index 98a4702..3dab6f8 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,4 +1,23 @@
-__pycache__
+# Python
+__pycache__/
+*.py[cod]
+*.pyo
+*.pyd
+.Python
+env/
+venv/
+ENV/
+
+# Streamlit
+.streamlit/
+
+# Environment variables
.env
-venv
-.venv
\ No newline at end of file
+
+# IDEs
+.vscode/
+.idea/
+
+# OS generated files
+.DS_Store
+Thumbs.db
\ No newline at end of file
diff --git a/README.md b/README.md
index 8ce90bb..502c276 100644
--- a/README.md
+++ b/README.md
@@ -14,9 +14,9 @@
- [ ] Add more providers
- [ ] Use something like LiteLLM to unify models code and avoid repeating code for each provider
-## Original G1 Example video
+## Example
-[Video Demo](https://github.com/user-attachments/assets/db2a221f-f8eb-48c3-b5a7-8399c6300243)
+
## Description
@@ -36,32 +36,12 @@ At each step, the LLM can choose to continue to another reasoning step, or provi
The reasoning ability of the LLM is therefore improved through combining Chain-of-Thought with the requirement to try multiple methods, explore alternative answers, question previous draft solutions, and consider the LLM’s limitations. This alone, without any training, is sufficient to achieve ~70% accuracy on the Strawberry problem (n=10, "How many Rs are in strawberry?"). Without prompting, Llama-3.1-70b had 0% accuracy and ChatGPT-4o had 30% accuracy.
-### Examples
+### Disclaimer
> [!IMPORTANT]
> multi1 is not perfect, but it can perform significantly better than LLMs out-of-the-box. From initial testing, multi1 accurately solves simple logic problems 60-80% of the time that usually stump LLMs. However, accuracy has yet to be formally evaluated. See examples below.
-##### How many Rs are in strawberry?
-Prompt: How many Rs are in strawberry?
-
-Result:
-
-
-
----
-
-Prompt: Which is larger, .9 or .11?
-
-Result:
-
-
-
-
-Prompt: In the context of Lie Group and Lie Algebra, let $R \in E$ be an irreducible root system. Show that then $E$ is an irreducible representation of the Weyl group $W$.
-
-
-
### Quickstart
diff --git a/__pycache__/api_handlers.cpython-310.pyc b/__pycache__/api_handlers.cpython-310.pyc
index 97b3881..275a2fc 100644
Binary files a/__pycache__/api_handlers.cpython-310.pyc and b/__pycache__/api_handlers.cpython-310.pyc differ
diff --git a/__pycache__/utils.cpython-310.pyc b/__pycache__/utils.cpython-310.pyc
index eed806d..b607340 100644
Binary files a/__pycache__/utils.cpython-310.pyc and b/__pycache__/utils.cpython-310.pyc differ
diff --git a/api_handlers.py b/app/api_handlers.py
similarity index 100%
rename from api_handlers.py
rename to app/api_handlers.py
diff --git a/app/config_menu.py b/app/config_menu.py
new file mode 100644
index 0000000..16d859b
--- /dev/null
+++ b/app/config_menu.py
@@ -0,0 +1,49 @@
+import streamlit as st
+import os
+from dotenv import load_dotenv, set_key
+
+def load_env_vars():
+ load_dotenv(os.path.join(os.path.dirname(__file__), "..", ".env"))
+ return {
+ 'OLLAMA_URL': os.getenv('OLLAMA_URL', 'http://localhost:11434'),
+ 'OLLAMA_MODEL': os.getenv('OLLAMA_MODEL', 'mistral'),
+ 'PERPLEXITY_API_KEY': os.getenv('PERPLEXITY_API_KEY', ''),
+ 'PERPLEXITY_MODEL': os.getenv('PERPLEXITY_MODEL', 'mistral-7b-instruct'),
+ 'GROQ_API_KEY': os.getenv('GROQ_API_KEY', ''),
+ 'GROQ_MODEL': os.getenv('GROQ_MODEL', 'mixtral-8x7b-32768')
+ }
+
+def save_env_vars(config):
+ env_path = os.path.join(os.path.dirname(__file__), "..", ".env")
+ for key, value in config.items():
+ set_key(env_path, key, value)
+
+def config_menu():
+ st.sidebar.markdown("## 🛠️ Configuration")
+
+ config = load_env_vars()
+
+ with st.sidebar.expander("Edit Configuration"):
+ new_config = {}
+ new_config['OLLAMA_URL'] = st.text_input("Ollama URL", value=config['OLLAMA_URL'])
+ new_config['OLLAMA_MODEL'] = st.text_input("Ollama Model", value=config['OLLAMA_MODEL'])
+ new_config['PERPLEXITY_API_KEY'] = st.text_input("Perplexity API Key", value=config['PERPLEXITY_API_KEY'], type="password")
+ new_config['PERPLEXITY_MODEL'] = st.text_input("Perplexity Model", value=config['PERPLEXITY_MODEL'])
+ new_config['GROQ_API_KEY'] = st.text_input("Groq API Key", value=config['GROQ_API_KEY'], type="password")
+ new_config['GROQ_MODEL'] = st.text_input("Groq Model", value=config['GROQ_MODEL'])
+
+ if st.button("Save Configuration"):
+ save_env_vars(new_config)
+ st.success("Configuration saved successfully!")
+
+ return config
+
+def display_config(backend, config):
+ st.sidebar.markdown("## 🛠️ Current Configuration")
+ if backend == "Ollama":
+ st.sidebar.markdown(f"- 🖥️ Ollama URL: `{config['OLLAMA_URL']}`")
+ st.sidebar.markdown(f"- 🤖 Ollama Model: `{config['OLLAMA_MODEL']}`")
+ elif backend == "Perplexity AI":
+ st.sidebar.markdown(f"- 🧠 Perplexity AI Model: `{config['PERPLEXITY_MODEL']}`")
+ else: # Groq
+ st.sidebar.markdown(f"- ⚡ Groq Model: `{config['GROQ_MODEL']}`")
diff --git a/app/main.py b/app/main.py
new file mode 100644
index 0000000..696268a
--- /dev/null
+++ b/app/main.py
@@ -0,0 +1,69 @@
+import streamlit as st
+from dotenv import load_dotenv
+from api_handlers import OllamaHandler, PerplexityHandler, GroqHandler
+from utils import generate_response
+from config_menu import config_menu, display_config
+import os
+
+# Load environment variables
+load_dotenv()
+
+def load_css():
+ with open(os.path.join(os.path.dirname(__file__), "..", "static", "styles.css")) as f:
+ st.markdown(f'', unsafe_allow_html=True)
+
+def setup_page():
+ st.set_page_config(page_title="multi1 - Unified AI Reasoning Chains", page_icon="🧠", layout="wide")
+ load_css()
+ st.markdown("""
+
+ 🧠 multi1 - Unified AI Reasoning Chains
+
+ """, unsafe_allow_html=True)
+ st.markdown("""
+
+ This app demonstrates AI reasoning chains using different backends: Ollama, Perplexity AI, and Groq.
+ Choose a backend and enter your query to see the step-by-step reasoning process.
+
+ """, unsafe_allow_html=True)
+
+def get_api_handler(backend, config):
+ if backend == "Ollama":
+ return OllamaHandler(config['OLLAMA_URL'], config['OLLAMA_MODEL'])
+ elif backend == "Perplexity AI":
+ return PerplexityHandler(config['PERPLEXITY_API_KEY'], config['PERPLEXITY_MODEL'])
+ else: # Groq
+ return GroqHandler(config['GROQ_API_KEY'], config['GROQ_MODEL'])
+
+def main():
+ setup_page()
+
+ st.sidebar.markdown('', unsafe_allow_html=True)
+ config = config_menu()
+
+ backend = st.sidebar.selectbox("Choose AI Backend", ["Ollama", "Perplexity AI", "Groq"])
+ display_config(backend, config)
+ api_handler = get_api_handler(backend, config)
+
+ user_query = st.text_input("💬 Enter your query:", placeholder="e.g., How many 'R's are in the word strawberry?")
+
+ if user_query:
+ st.write("🔍 Generating response...")
+ response_container = st.empty()
+ time_container = st.empty()
+
+ for steps, total_thinking_time in generate_response(user_query, api_handler):
+ with response_container.container():
+ for title, content, _ in steps:
+ if title.startswith("Final Answer"):
+ st.markdown(f'🎯 {title}
', unsafe_allow_html=True)
+ st.markdown(f'{content}
', unsafe_allow_html=True)
+ else:
+ with st.expander(f"📝 {title}", expanded=True):
+ st.markdown(f'{content}
', unsafe_allow_html=True)
+
+ if total_thinking_time is not None:
+ time_container.markdown(f'⏱️ Total thinking time: {total_thinking_time:.2f} seconds
', unsafe_allow_html=True)
+
+if __name__ == "__main__":
+ main()
\ No newline at end of file
diff --git a/utils.py b/app/utils.py
similarity index 100%
rename from utils.py
rename to app/utils.py
diff --git a/examples/lie.1.png b/examples/lie.1.png
deleted file mode 100644
index f990d3d..0000000
Binary files a/examples/lie.1.png and /dev/null differ
diff --git a/examples/math.png b/examples/math.png
deleted file mode 100644
index a20f324..0000000
Binary files a/examples/math.png and /dev/null differ
diff --git a/examples/maths.png b/examples/maths.png
new file mode 100644
index 0000000..cd8f6e9
Binary files /dev/null and b/examples/maths.png differ
diff --git a/examples/strawberry.png b/examples/strawberry.png
deleted file mode 100644
index a8f677b..0000000
Binary files a/examples/strawberry.png and /dev/null differ
diff --git a/main.py b/main.py
deleted file mode 100644
index fb21acc..0000000
--- a/main.py
+++ /dev/null
@@ -1,71 +0,0 @@
-import streamlit as st
-from dotenv import load_dotenv
-from api_handlers import OllamaHandler, PerplexityHandler, GroqHandler
-from utils import generate_response, load_env_vars
-
-# Load environment variables and configuration
-load_dotenv()
-config = load_env_vars()
-
-def setup_page():
- st.set_page_config(page_title="multi1 - Unified AI Reasoning Chains", page_icon="🧠", layout="wide")
- st.markdown("""
-
- 🧠 multi1 - Unified AI Reasoning Chains
-
- """, unsafe_allow_html=True)
- st.markdown("""
-
- This app demonstrates AI reasoning chains using different backends: Ollama, Perplexity AI, and Groq.
- Choose a backend and enter your query to see the step-by-step reasoning process.
-
- """, unsafe_allow_html=True)
-
-def get_api_handler(backend):
- if backend == "Ollama":
- return OllamaHandler(config['OLLAMA_URL'], config['OLLAMA_MODEL'])
- elif backend == "Perplexity AI":
- return PerplexityHandler(config['PERPLEXITY_API_KEY'], config['PERPLEXITY_MODEL'])
- else: # Groq
- return GroqHandler()
-
-def display_config(backend):
- st.sidebar.markdown("## 🛠️ Current Configuration")
- if backend == "Ollama":
- st.sidebar.markdown(f"- 🖥️ Ollama URL: `{config['OLLAMA_URL']}`")
- st.sidebar.markdown(f"- 🤖 Ollama Model: `{config['OLLAMA_MODEL']}`")
- elif backend == "Perplexity AI":
- st.sidebar.markdown(f"- 🧠 Perplexity AI Model: `{config['PERPLEXITY_MODEL']}`")
- else: # Groq
- st.sidebar.markdown("- ⚡ Using Groq API")
-
-def main():
- setup_page()
-
- st.sidebar.markdown("⚙️ Settings
", unsafe_allow_html=True)
- backend = st.sidebar.selectbox("Choose AI Backend", ["Ollama", "Perplexity AI", "Groq"])
- display_config(backend)
- api_handler = get_api_handler(backend)
-
- user_query = st.text_input("💬 Enter your query:", placeholder="e.g., How many 'R's are in the word strawberry?")
-
- if user_query:
- st.write("🔍 Generating response...")
- response_container = st.empty()
- time_container = st.empty()
-
- for steps, total_thinking_time in generate_response(user_query, api_handler):
- with response_container.container():
- for title, content, _ in steps:
- if title.startswith("Final Answer"):
- st.markdown(f"🎯 {title}
", unsafe_allow_html=True)
- st.markdown(f"{content}
", unsafe_allow_html=True)
- else:
- with st.expander(f"📝 {title}", expanded=True):
- st.markdown(f"{content}
", unsafe_allow_html=True)
-
- if total_thinking_time is not None:
- time_container.markdown(f"⏱️ Total thinking time: {total_thinking_time:.2f} seconds
", unsafe_allow_html=True)
-
-if __name__ == "__main__":
- main()
\ No newline at end of file
diff --git a/static/styles.css b/static/styles.css
new file mode 100644
index 0000000..d2ee572
--- /dev/null
+++ b/static/styles.css
@@ -0,0 +1,29 @@
+body {
+ font-family: -apple-system, BlinkMacSystemFont, sans-serif;
+}
+
+h1, h2, h3, h4, h5, h6 {
+ font-family: -apple-system, BlinkMacSystemFont, sans-serif;
+}
+
+.main-title {
+ text-align: center;
+}
+
+.main-description {
+ text-align: center;
+ font-size: 1.1em;
+}
+
+.sidebar-title {
+ font-family: -apple-system, BlinkMacSystemFont, sans-serif;
+}
+
+.expander-title {
+ font-family: -apple-system, BlinkMacSystemFont, sans-serif;
+}
+
+.thinking-time {
+ font-family: -apple-system, BlinkMacSystemFont, sans-serif;
+ font-weight: bold;
+}