{"id":8750,"date":"2024-01-17T06:00:07","date_gmt":"2024-01-17T14:00:07","guid":{"rendered":"https:\/\/live-cometml.pantheonsite.io\/?p=8750"},"modified":"2025-04-24T17:03:31","modified_gmt":"2025-04-24T17:03:31","slug":"exploring-the-power-of-llama-2-using-streamlit","status":"publish","type":"post","link":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit\/","title":{"rendered":"Exploring the Power of Llama 2 Using Streamlit"},"content":{"rendered":"\n
\"llama\"\/
Photo by Trace Hudson<\/a><\/figcaption><\/figure>\n\n\n\n

Introduction<\/h4>\n\n\n\n

2023 has been significant for large language models. Many advancements have been made since ChatGPT, including open-source and licensed models. The recent release of Llama 2 by Meta and Microsoft has taken the AI world by storm. An open-source model that is just as good as GPT 3.5 or even GPT 4? We\u2019re yet to know.<\/p>\n\n\n\n

In this article, we will explore the power of Llama 2 using Streamlit<\/a>.<\/p>\n\n\n\n

Understanding Llama v2<\/h4>\n\n\n\n

Llama 2, a natural language processing library from Meta, consists of pre-trained and fine-tuned large language models (LLMs) whose scale ranges from 7 billion to 70 billion parameters. The Llama 2-Chat is optimized to support human conversation use cases. It is a suitable alternative for closed models since it outperforms the open-source chat models on most tested benchmarks. We\u2019ll explore how Streamlit provides an intuitive user interface, and Llama v2 empowers the chatbot with sophisticated language understanding capabilities, resulting in a powerful conversational AI application.<\/p>\n\n\n\n

Setting Up Streamlit<\/h4>\n\n\n\n

The first step to setting up a Streamlit app is to create a virtual environment.<\/p>\n\n\n\n

We will use the Pipenv <\/strong>library as follows:<\/p>\n\n\n\n

pipenv shell<\/span><\/pre>\n\n\n\n
Next, we will install the necessary libraries to build our chatbot.<\/p>\n\n\n\n
pipenv install streamlit replicate<\/span><\/pre>\n\n\n\n
Streamlit <\/strong>is an open-source Python framework used to deploy data science apps.<\/p>\n\n\n\n
Replicate <\/strong>is a cloud platform that hosts large machine learning models for easy deployment.<\/p>\n\n\n\n
Accessing Llama 2 API Token<\/h4>\n\n\n\n
Before you access Replicate\u2019s token key, you must register an account on Replicate<\/a>.<\/p>\n\n\n\n
\n
Note<\/strong>: You must have a GitHub<\/a> account to sign in to Replicate.<\/p>\n<\/blockquote>\n\n\n\n
Search for Llama 2 chat on the Replicate dashboard.<\/p>\n\n\n\n
\"Llama
Replicate Dashboard<\/figcaption><\/figure>\n\n\n\n
 <\/p>\n\n\n\n
Click on the llama-2\u201370b-chat<\/strong> model to view the Llama 2 API endpoints.<\/p>\n\n\n\n
Click on the API button on the llama-2\u201370b-chat<\/strong> model\u2019s navbar.<\/p>\n\n\n\n
Next, on the right side of the page, click on the Python <\/strong>button to access the API token for Python Applications.<\/p>\n\n\n\n
\"screenshot<\/figure>\n\n\n\n
Copy the REPLICATE_API_TOKEN <\/strong>and keep it safe for future use.<\/p>\n\n\n\n
Building A Chatbot<\/h4>\n\n\n\n
First, create a Python file called chatbot.py<\/em> and .env <\/em>file. We will write our code on the chatbot.py and store our secret keys and API tokens in the .env<\/em> file.<\/p>\n\n\n\n
On the chatbot.py file, import the libraries as follows.<\/p>\n\n\n\n
import<\/span> streamlit as<\/span> st\nimport<\/span> os\nimport<\/span> replicate<\/span><\/pre>\n\n\n\n
Next, set the global variables of the llama-2\u201370b-chat<\/strong> model.<\/p>\n\n\n\n
# Global variables<\/span>\nREPLICATE_API_TOKEN = os.environ.get<\/span>('REPLICATE_API_TOKEN'<\/span>, default<\/span>=''<\/span>)\nREPLICATE_MODEL_ENDPOINTS = {\n    'LLaMA2-7B'<\/span>: os.environ.get<\/span>('REPLICATE_MODEL_ENDPOINT7B'<\/span>, default<\/span>=''<\/span>),\n    'LLaMA2-13B'<\/span>: os.environ.get<\/span>('REPLICATE_MODEL_ENDPOINT13B'<\/span>, default<\/span>=''<\/span>),\n    'LLaMA2-70B'<\/span>: os.environ.get<\/span>('REPLICATE_MODEL_ENDPOINT70B'<\/span>, default<\/span>=''<\/span>)\n}<\/span><\/pre>\n\n\n\n
On the .env<\/em> file, add the Replicate token and model endpoints in the following format:<\/p>\n\n\n\n
# \ud83d\udcc1 .env -----<\/span>\nREPLICATE_API_TOKEN<\/span>='Your_Replicate_Token'<\/span>\nREPLICATE_MODEL_ENDPOINT7B<\/span>='a16z-infra\/llama7b-v2-chat:4f0a4744c7295c024a1de15e1a63c880d3da035fa1f49bfd344fe076074c8eea'<\/span>\nREPLICATE_MODEL_ENDPOINT13B<\/span>='a16z-infra\/llama13b-v2-chat:df7690f1994d94e96ad9d568eac121aecf50684a0b0963b25a41cc40061269e5'<\/span>\nREPLICATE_MODEL_ENDPOINT70B<\/span>='replicate\/llama70b-v2-chat:e951f18578850b652510200860fc4ea62b3b16fac280f83ff32282f87bbd2e48'<\/span><\/span><\/pre>\n\n\n\n
Create a pre-prompt for the Llama 2 model. We will write a simple prompt to set the model as an assistant in this case.<\/p>\n\n\n\n
# Set Pre-propmt<\/span>\nPRE_PROMPT<\/span> = \"You are a helpful assistant. You do not respond as 'User' or pretend to be 'User'. You only respond once as Assistant.\"<\/span><\/span><\/pre>\n\n\n\n
Set up the page configuration for our chatbot as follows:<\/p>\n\n\n\n
# Set initial page configuration<\/span>\nst.set_page_config(page_title=\"LLaMA2 Chatbot\"<\/span>, page_icon=\":left_speech_bubble:\"<\/span>, layout=\"wide\"<\/span>)<\/span><\/pre>\n\n\n\n
Write a function to handle the chatbot functionality.<\/p>\n\n\n\n
def render_app():\n    # Set up containers\n    response_container = st.container<\/span>()\n    container = st.container<\/span>()\n\n    # Set up Session State variables\n    st.session_state.setdefault<\/span>('chat_dialogue'<\/span>, [])\n    st.session_state.setdefault<\/span>('llm'<\/span>, REPLICATE_MODEL_ENDPOINTS['LLaMA2-70B'<\/span>])\n    st.session_state.setdefault<\/span>('temperature'<\/span>, 0.1<\/span>)\n    st.session_state.setdefault<\/span>('top_p'<\/span>, 0.9<\/span>)\n    st.session_state.setdefault<\/span>('max_seq_len'<\/span>, 512<\/span>)\n    st.session_state.setdefault<\/span>('pre_prompt'<\/span>, PRE_PROMPT)\n    st.session_state.setdefault<\/span>('string_dialogue'<\/span>, ''<\/span>)\n\n\n    # Display chat history\n    for message in st.session_state.chat_dialogue:\n        with st.chat_message<\/span>(message[\"role\"<\/span>]):\n            st.markdown<\/span>(message[\"content\"<\/span>])\n\n    # User input\n    if prompt := st.chat_input<\/span>(\"Ask a question here to LLaMA2\"<\/span>):\n        st.session_state.chat_dialogue.append<\/span>({\"role\": \"user\"<\/span>, \"content\"<\/span>: prompt})\n        with st.chat_message<\/span>(\"user\"):\n            st.markdown<\/span>(prompt)\n\n        # Assistant response\n        with st.chat_message<\/span>(\"assistant\"<\/span>):\n            message_placeholder = st.empty<\/span>()\n            full_response = \"\"<\/span>\n            string_dialogue = st.session_state['pre_prompt'<\/span>]\n            for dict_message in st.session_state.chat_dialogue:\n                speaker = \"User\"<\/span> if dict_message[\"role\"<\/span>] == \"user\"<\/span> else \"Assistant\"<\/span>\n                string_dialogue += f\"{speaker}: {dict_message['content']}\\n\\n\"<\/span>\n            output = debounce_replicate_run<\/span>(\n                st.session_state['llm'<\/span>],\n                string_dialogue + \"Assistant: \"<\/span>,\n                st.session_state['max_seq_len'<\/span>],\n                st.session_state['temperature'<\/span>],\n                st.session_state['top_p'<\/span>],\n                REPLICATE_API_TOKEN\n            )\n            for item in output:\n                full_response += item\n                message_placeholder.markdown<\/span>(full_response + \"\u258c\"<\/span>)\n            message_placeholder.markdown<\/span>(full_response)\n        st.session_state.chat_dialogue.append<\/span>({\"role\": \"assistant\"<\/span>, \"content\"<\/span>: full_response})<\/span><\/pre>\n\n\n\n
Page UI Configurations<\/h4>\n\n\n\n
Within the render_app()<\/em> function, immediately after the session state variables, add the following lines of code to enable users to customize the chatbot with their desired Llama model and custom prompt:<\/p>\n\n\n\n
# Set up left sidebar<\/span>\n    st.sidebar.header(\"LLaMA2 Chatbot\"<\/span>)\n\n     #container for the chat history<\/span>\n    response_container = st.container()\n    #container for the user's text input<\/span>\n    container = st.container()\n    #Set up\/Initialize Session State variables:<\/span>\n    if<\/span> 'chat_dialogue'<\/span> not<\/span> in<\/span> st.session_state:\n        st.session_state['chat_dialogue'<\/span>] = []\n    if<\/span> 'llm'<\/span> not<\/span> in<\/span> st.session_state:\n        #st.session_state['llm'] = REPLICATE_MODEL_ENDPOINT13B<\/span>\n        st.session_state['llm'<\/span>] = 'LLaMA2-70B'<\/span>\n    if<\/span> 'temperature'<\/span> not<\/span> in<\/span> st.session_state:\n        st.session_state['temperature'<\/span>] = 0.1<\/span>\n    if<\/span> 'top_p'<\/span> not<\/span> in<\/span> st.session_state:\n        st.session_state['top_p'<\/span>] = 0.9<\/span>\n    if<\/span> 'max_seq_len'<\/span> not<\/span> in<\/span> st.session_state:\n        st.session_state['max_seq_len'<\/span>] = 512<\/span>\n    if<\/span> 'pre_prompt'<\/span> not<\/span> in<\/span> st.session_state:\n        st.session_state['pre_prompt'<\/span>] = PRE_PROMPT\n    if<\/span> 'string_dialogue'<\/span> not<\/span> in<\/span> st.session_state:\n        st.session_state['string_dialogue'<\/span>] = ''<\/span>\n\n    #Dropdown menu to select the model edpoint:<\/span>\n    selected_option = st.sidebar.selectbox('Choose a LLaMA2 model:'<\/span>, ['LLaMA2-70B'<\/span>, 'LLaMA2-13B'<\/span>, 'LLaMA2-7B'<\/span>], key='model'<\/span>)\n    if<\/span> selected_option == 'LLaMA2-7B'<\/span>:\n        st.session_state['llm'<\/span>] = 'LLaMA2-7B'<\/span>\n    elif<\/span> selected_option == 'LLaMA2-13B'<\/span>:\n        st.session_state['llm'<\/span>] = 'LLaMA2-13B'<\/span>\n    else<\/span>:\n        st.session_state['llm'<\/span>] = 'LLaMA2-70B'<\/span>\n    #Model hyper parameters:<\/span>\n    st.session_state['temperature'<\/span>] = st.sidebar.slider('Temperature:'<\/span>, min_value=0.01<\/span>, max_value=5.0<\/span>, value=0.1<\/span>, step=0.01<\/span>)\n    st.session_state['top_p'<\/span>] = st.sidebar.slider('Top P:'<\/span>, min_value=0.01<\/span>, max_value=1.0<\/span>, value=0.9<\/span>, step=0.01<\/span>)\n    st.session_state['max_seq_len'<\/span>] = st.sidebar.slider('Max Sequence Length:'<\/span>, min_value=64<\/span>, max_value=4096<\/span>, value=2048<\/span>, step=8<\/span>)\n\n    NEW_P = st.sidebar.text_area('Prompt before the chat starts. Edit here if desired:'<\/span>, PRE_PROMPT, height=60<\/span>)\n    if<\/span> NEW_P != PRE_PROMPT and<\/span> NEW_P != \"\"<\/span> and<\/span> NEW_P != None<\/span>:\n        st.session_state['pre_prompt'<\/span>] = NEW_P + \"\\n\\n\"<\/span>\n    else<\/span>:\n        st.session_state['pre_prompt'<\/span>] = PRE_PROMPT\n\n    btn_col1, btn_col2 = st.sidebar.columns(2<\/span>)<\/span><\/pre>\n\n\n\n
Next, call the render_app()<\/em> function:<\/p>\n\n\n\n
def main<\/span>():\n    render_app<\/span>()\n\nif<\/span> __name__<\/span> == \"__main__\"<\/span>:\n    main()<\/span><\/pre>\n\n\n\n
Once done, add a utils.py<\/em> file and import it into your chatbot.py<\/em> file.<\/p>\n\n\n\n
On the utils.py file, create a debounce replicate run function that ensures the model usage does not result in poor performance, double activations, or user frustrations.<\/p>\n\n\n\n
import<\/span> replicate\nimport<\/span> time\n\n# Initialize debounce variables<\/span>\nlast_call_time = 0<\/span>\ndebounce_interval = 2<\/span>  # Set the debounce interval (in seconds) to your desired value<\/span>\n\ndef<\/span> debounce_replicate_run<\/span>(llm, prompt, max_len, temperature, top_p, API_TOKEN<\/span>):\n    global<\/span> last_call_time\n    print<\/span>(\"last call time: \"<\/span>, last_call_time)\n\n    # Get the current time<\/span>\n    current_time = time.time()\n\n    # Calculate the time elapsed since the last call<\/span>\n    elapsed_time = current_time - last_call_time\n\n    # Check if the elapsed time is less than the debounce interval<\/span>\n    if<\/span> elapsed_time < debounce_interval:\n        print<\/span>(\"Debouncing\"<\/span>)\n        return<\/span> \"Hello! You are sending requests too fast. Please wait a few seconds before sending another request.\"<\/span>\n\n\n    # Update the last call time to the current time<\/span>\n    last_call_time = time.time()\n\n    output = replicate.run(llm, input<\/span>={\"prompt\"<\/span>: prompt + \"Assistant: \"<\/span>, \"max_length\"<\/span>: max_len, \"temperature\"<\/span>: temperature, \"top_p\"<\/span>: top_p, \"repetition_penalty\"<\/span>: 1<\/span>}, api_token=API_TOKEN)\n    return<\/span> output\n\n\n<\/span><\/pre>\n\n\n\n
In this code, the debounce_replicate_run<\/em> function implements a debounce mechanism to help prevent frequent and excessive API queries from a user\u2019s input.<\/p>\n\n\n\n
The function makes API calls using the Replicate <\/strong>library and manages time intervals using the time <\/strong>module. The function accepts several API call-related parameters, including model, prompt, maximum length, temperature, and API token<\/em>. The debounce process includes comparing the time from the most recent API call to a predetermined interval.<\/p>\n\n\n\n
The function provides a message indicating that requests are being sent too quickly and advises the user to wait if the time elapsed is less than the debounce interval. The function changes the last call time and makes the API request utilizing the debounce interval if it has.<\/p>\n\n\n\n
Back on the chatbot.py file, import the debounce_replicate_run<\/em> function from utils as follows:<\/p>\n\n\n\n
from<\/span> utils import<\/span> debounce_replicate_run<\/span><\/pre>\n\n\n\n
After that, you can now run your application using the command below:<\/p>\n\n\n\n
streamlit run chatbot.py<\/span><\/span><\/pre>\n\n\n\n
Expected Results:<\/p>\n\n\n\n
\"Llama
Llamav2 and Streamlit app screenshot <\/figcaption><\/figure>\n\n\n\n
Conclusion<\/h4>\n\n\n\n
From the tutorial, you have learned how to explore the power of Llama2 using Streamlit. Compared to the popular closed-source model, GPT-3.5, which is based on data up to 2021, Llama2 is trained on data up to December 2022. This makes the model a capable rival to some of the closed models in the market, advancing the development of AI applications at lower costs.<\/p>\n","protected":false},"excerpt":{"rendered":"
Introduction 2023 has been significant for large language models. Many advancements have been made since ChatGPT, including open-source and licensed models. The recent release of Llama 2 by Meta and Microsoft has taken the AI world by storm. An open-source model that is just as good as GPT 3.5 or even GPT 4? We\u2019re yet […]<\/p>\n","protected":false},"author":52,"featured_media":8806,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"customer_name":"","customer_description":"","customer_industry":"","customer_technologies":"","customer_logo":"","footnotes":""},"categories":[7],"tags":[],"coauthors":[154],"class_list":["post-8750","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tutorials"],"yoast_head":"\nExploring the Power of Llama 2 Using Streamlit - Comet<\/title>\n<meta name=\"description\" content=\"In this blog article, learn how to explore the power of Llama 2 using Streamlit in a step by step tutorial.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Exploring the Power of Llama 2 Using Streamlit\" \/>\n<meta property=\"og:description\" content=\"In this blog article, learn how to explore the power of Llama 2 using Streamlit in a step by step tutorial.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit\" \/>\n<meta property=\"og:site_name\" content=\"Comet\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/cometdotml\" \/>\n<meta property=\"article:published_time\" content=\"2024-01-17T14:00:07+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2025-04-24T17:03:31+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"800\" \/>\n\t<meta property=\"og:image:height\" content=\"575\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Adhing'a Fredrick\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:image\" content=\"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg\" \/>\n<meta name=\"twitter:creator\" content=\"@Cometml\" \/>\n<meta name=\"twitter:site\" content=\"@Cometml\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Adhing'a Fredrick\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"7 minutes\" \/>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Exploring the Power of Llama 2 Using Streamlit - Comet","description":"In this blog article, learn how to explore the power of Llama 2 using Streamlit in a step by step tutorial.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit","og_locale":"en_US","og_type":"article","og_title":"Exploring the Power of Llama 2 Using Streamlit","og_description":"In this blog article, learn how to explore the power of Llama 2 using Streamlit in a step by step tutorial.","og_url":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit","og_site_name":"Comet","article_publisher":"https:\/\/www.facebook.com\/cometdotml","article_published_time":"2024-01-17T14:00:07+00:00","article_modified_time":"2025-04-24T17:03:31+00:00","og_image":[{"width":800,"height":575,"url":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg","type":"image\/jpeg"}],"author":"Adhing'a Fredrick","twitter_card":"summary_large_image","twitter_image":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg","twitter_creator":"@Cometml","twitter_site":"@Cometml","twitter_misc":{"Written by":"Adhing'a Fredrick","Est. reading time":"7 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit#article","isPartOf":{"@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit\/"},"author":{"name":"Adhing'a Fredrick","@id":"https:\/\/www.comet.com\/site\/#\/schema\/person\/2f5012b38311f83e8165100848778d19"},"headline":"Exploring the Power of Llama 2 Using Streamlit","datePublished":"2024-01-17T14:00:07+00:00","dateModified":"2025-04-24T17:03:31+00:00","mainEntityOfPage":{"@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit\/"},"wordCount":747,"publisher":{"@id":"https:\/\/www.comet.com\/site\/#organization"},"image":{"@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit#primaryimage"},"thumbnailUrl":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg","articleSection":["Tutorials"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit\/","url":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit","name":"Exploring the Power of Llama 2 Using Streamlit - Comet","isPartOf":{"@id":"https:\/\/www.comet.com\/site\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit#primaryimage"},"image":{"@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit#primaryimage"},"thumbnailUrl":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg","datePublished":"2024-01-17T14:00:07+00:00","dateModified":"2025-04-24T17:03:31+00:00","description":"In this blog article, learn how to explore the power of Llama 2 using Streamlit in a step by step tutorial.","breadcrumb":{"@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit#primaryimage","url":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg","contentUrl":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2024\/01\/Llama.jpg","width":800,"height":575,"caption":"llama"},{"@type":"BreadcrumbList","@id":"https:\/\/www.comet.com\/site\/blog\/exploring-the-power-of-llama-2-using-streamlit#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.comet.com\/site\/"},{"@type":"ListItem","position":2,"name":"Exploring the Power of Llama 2 Using Streamlit"}]},{"@type":"WebSite","@id":"https:\/\/www.comet.com\/site\/#website","url":"https:\/\/www.comet.com\/site\/","name":"Comet","description":"Build Better Models Faster","publisher":{"@id":"https:\/\/www.comet.com\/site\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.comet.com\/site\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/www.comet.com\/site\/#organization","name":"Comet ML, Inc.","alternateName":"Comet","url":"https:\/\/www.comet.com\/site\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.comet.com\/site\/#\/schema\/logo\/image\/","url":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2025\/01\/logo_comet_square.png","contentUrl":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2025\/01\/logo_comet_square.png","width":310,"height":310,"caption":"Comet ML, Inc."},"image":{"@id":"https:\/\/www.comet.com\/site\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/cometdotml","https:\/\/x.com\/Cometml","https:\/\/www.youtube.com\/channel\/UCmN63HKvfXSCS-UwVwmK8Hw"]},{"@type":"Person","@id":"https:\/\/www.comet.com\/site\/#\/schema\/person\/2f5012b38311f83e8165100848778d19","name":"Adhing'a Fredrick","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.comet.com\/site\/#\/schema\/person\/image\/8f4b6487c0f3b99fbc3e482d4b376924","url":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2023\/08\/1652714258953-96x96.jpg","contentUrl":"https:\/\/www.comet.com\/site\/wp-content\/uploads\/2023\/08\/1652714258953-96x96.jpg","caption":"Adhing'a Fredrick"},"url":"https:\/\/www.comet.com\/site\/blog\/author\/adhingafredrickgmail-com\/"}]}},"_links":{"self":[{"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/posts\/8750","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/users\/52"}],"replies":[{"embeddable":true,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/comments?post=8750"}],"version-history":[{"count":1,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/posts\/8750\/revisions"}],"predecessor-version":[{"id":15403,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/posts\/8750\/revisions\/15403"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/media\/8806"}],"wp:attachment":[{"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/media?parent=8750"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/categories?post=8750"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/tags?post=8750"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.comet.com\/site\/wp-json\/wp\/v2\/coauthors?post=8750"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}