File size: 5,400 Bytes
2a88707
 
34d0739
f68537f
a05fede
34d0739
a05fede
34d0739
 
92d7276
2a88707
34d0739
442574e
 
45963cd
 
442574e
 
 
2a88707
 
 
f068886
92d7276
2a88707
34d0739
174910b
ae1353b
 
 
174910b
622f160
ae1353b
174910b
2a88707
34d0739
 
 
 
e88412a
2a88707
 
 
e88412a
 
ae1353b
 
e88412a
77ae47f
ae1353b
622f160
77ae47f
622f160
174910b
 
 
622f160
 
 
 
174910b
 
622f160
174910b
2a88707
65dc20f
2a88707
174910b
622f160
174910b
 
 
622f160
ae1353b
 
622f160
ae1353b
174910b
622f160
174910b
 
622f160
ae1353b
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
import streamlit as st
import cohere
import os
import base64

st.set_page_config(page_title="Cohere Chat", layout="wide")

AI_PFP = "media/pfps/cohere-pfp.png"
USER_PFP = "media/pfps/user-pfp.jpg"
BANNER = "media/banner.png"

model_info = {
    "c4ai-aya-expanse-8b": {"description": "Aya Expanse is a highly performant 8B multilingual model, designed to rival monolingual performance through innovations in instruction tuning with data arbitrage, preference training, and model merging. Serves 23 languages.", "context": "4K", "output": "4K"},
    "c4ai-aya-expanse-32b": {"description": "Aya Expanse is a highly performant 32B multilingual model, designed to rival monolingual performance through innovations in instruction tuning with data arbitrage, preference training, and model merging. Serves 23 languages.", "context": "128K", "output": "4K"},
    "c4ai-aya-vision-8b": {"description": "Aya Vision is a state-of-the-art multimodal model excelling at a variety of critical benchmarks for language, text, and image capabilities. This 8 billion parameter variant is focused on low latency and best-in-class performance.", "context": "16K", "output": "4K"},
    "c4ai-aya-vision-32b": {"description": "Aya Vision is a state-of-the-art multimodal model excelling at a variety of critical benchmarks for language, text, and image capabilities. Serves 23 languages. This 32 billion parameter variant is focused on state-of-art multilingual performance.", "context": "16k", "output": "4K"},
    "command-a-03-2025": {"description": "Command A is our most performant model to date, excelling at tool use, agents, retrieval augmented generation (RAG), and multilingual use cases. Command A has a context length of 256K, only requires two GPUs to run, and has 150% higher throughput compared to Command R+ 08-2024.", "context": "256K", "output": "8K"},
    "command-r7b-12-2024": {"description": "command-r7b-12-2024 is a small, fast update delivered in December 2024. It excels at RAG, tool use, agents, and similar tasks requiring complex reasoning and multiple steps.", "context": "128K", "output": "4K"},
    "command-r-plus-04-2024": {"description": "Command R+ is an instruction-following conversational model that performs language tasks at a higher quality, more reliably, and with a longer context than previous models. It is best suited for complex RAG workflows and multi-step tool use.", "context": "128K", "output": "4K"},
}

with st.sidebar:
    st.image(BANNER, use_container_width=True)
    st.title("Settings")
    api_key = st.text_input("Cohere API Key", type="password")
    selected_model = st.selectbox("Model", options=list(model_info.keys()))
    if selected_model.startswith("c4ai-aya-vision"):
        uploaded_file = st.file_uploader("Upload image", type=["png", "jpg", "jpeg"])
        if uploaded_file:
            st.session_state.image_data = uploaded_file.read()
    if st.button("Clear Chat"):
        st.session_state.clear()
        st.session_state.image_data = None
        st.rerun()
    st.divider()
    st.subheader(selected_model)
    st.markdown(model_info[selected_model]["description"])
    st.caption(f"Context: {model_info[selected_model]['context']}")
    st.caption(f"Output: {model_info[selected_model]['output']}")
    st.markdown("Powered by Cohere's API")

if "messages" not in st.session_state:
    st.session_state.messages = []
if "first_message_sent" not in st.session_state:
    st.session_state.first_message_sent = False
if "image_data" not in st.session_state:
    st.session_state.image_data = None

if not st.session_state.first_message_sent:
    st.markdown("<h1 style='text-align:center;color:#4a4a4a;margin-top:100px;'>How can Cohere help you today?</h1>", unsafe_allow_html=True)

for msg in st.session_state.messages:
    with st.chat_message(msg["role"], avatar=USER_PFP if msg["role"] == "user" else AI_PFP):
        content = msg["content"]
        if isinstance(content, list):
            for item in content:
                if getattr(item, 'type', None) == "text":
                    st.markdown(item.text)
                elif getattr(item, 'type', None) == "image_url":
                    st.image(item.image_url.url)
        else:
            st.markdown(content)

if prompt := st.chat_input("Message..."):
    if not api_key:
        st.error("API key required! Please create one at https://dashboard.cohere.com/")
        st.stop()
    st.session_state.first_message_sent = True
    st.session_state.messages.append({"role": "user", "content": prompt})
    with st.chat_message("user", avatar=USER_PFP):
        st.markdown(prompt)
    co = cohere.ClientV2(api_key)
    content = [{"type": "text", "text": prompt}]
    if selected_model.startswith("c4ai-aya-vision") and st.session_state.image_data:
        data_url = "data:image/jpeg;base64," + base64.b64encode(st.session_state.image_data).decode('utf-8')
        content.append({"type": "image_url", "image_url": {"url": data_url}})
    response = co.chat(model=selected_model, messages=[*st.session_state.messages, {"role": "user", "content": content}], temperature=0.3)
    items = response.message.content
    reply = "".join(getattr(i, 'text', '') for i in items)
    with st.chat_message("assistant", avatar=AI_PFP):
        st.markdown(reply)
    st.session_state.messages.append({"role": "assistant", "content": items})
    st.session_state.image_data = None