Update

app.py

@@ -0,0 +1,70 @@
+from matplotlib.pyplot import get
+from matplotlib.style import available
+import streamlit as st
+import numpy as np
+import pandas as pd
+import streamlit.components.v1 as components
+
+from millify import millify
+
+from utils.utils_display import get_current_date, get_json_from_date, get_available_dates, render_st_from_chapter_number, get_current_global_step
+from utils.constants import preface_disclaimer
+
+st.set_page_config(page_title="Bloom Book",layout='wide')
+
+BATCH_SIZE=2048
+SEQ_LENGTH=2048
+
+curr_date = get_current_date()
+
+# set_png_as_page_bg("data/image/bloom-book-bg.png") #
+st.markdown("<h1 style='text-align: center; color: grey;'>📖 BLOOM Book 📖 </h1>", unsafe_allow_html=True)
+
+available_dates = get_available_dates()
+available_chapters = ("Preface", ) + tuple(available_dates)
+
+st.sidebar.image(
+    "https://assets.website-files.com/6139f3cdcbbff3a68486761d/613cd8997b270da063e230c5_Tekengebied%201-p-2000.png",
+    use_column_width=True
+)
+
+st.sidebar.title(
+    "Chapters browser"
+)
+
+st.sidebar.markdown(
+    "You can freely browse the different chapters - ie example prompts from different people - and see the results."
+)
+
+selected_date = st.sidebar.selectbox(
+    "Please select the chapter you want to read:",
+    available_chapters
+)
+
+if selected_date != "Preface":
+    current_global_step = get_current_global_step(selected_date)
+    seen_tokens = BATCH_SIZE * SEQ_LENGTH * current_global_step
+    st.markdown("<h2 style='text-align: center; color: grey;'> Chapter {} </h2>".format(selected_date), unsafe_allow_html=True)
+    st.markdown("<h3 style='text-align: center; color: grey;'> Global step: {} - Seen tokens: {} </h3>".format(current_global_step, millify(seen_tokens)), unsafe_allow_html=True)
+    st.markdown("<h5 style='text-align: center; color: grey;'> Click into the text cards to visualize the answers </h5>", unsafe_allow_html=True)
+
+    selected_format = st.sidebar.selectbox('Visualize as:', ["HTML","JSON"])
+    suffixes = ["greedy", "nucleus"]
+
+    if selected_format == "HTML":
+        user_input = st.sidebar.text_input("Search for a specific prompt: ", "")
+        render_st_from_chapter_number(selected_date, suffixes, user_input)
+    elif selected_format == "JSON":
+        suffix = st.sidebar.selectbox('Decoding strategy:', ["greedy","nucleus"])
+        json_output = get_json_from_date(selected_date, suffix)
+        st.json(json_output)
+else:
+    st.markdown("<h3 style='text-align: center; color: grey;'> Welcome to the <i> BLOOM Book </i>. Here you can read generations from the main model based on prompts provided by the community. </h3> ", unsafe_allow_html=True)
+    st.markdown("""<h3 style='text-align: center; color: grey;'> Follow the main model's training <a href='https://huggingface.co/bigscience/tr11-176B-ml-logs' target="_blank"> here </a> </h3> """, unsafe_allow_html=True)
+    st.markdown("""<h3 style='text-align: center; color: grey;'> Try your own prompts? Check the <a href='https://forms.gle/2L7jkZt8MS8VDy2ZA' target="_blank"> Google Form </a> </h3> """, unsafe_allow_html=True)
+    st.markdown("{}".format(preface_disclaimer), unsafe_allow_html=True)
+    final_html =""" """ #TODO: add preface
+    chapter = components.html(
+        final_html,
+        height=600,
+    )