Spaces:
Runtime error
Runtime error
add comments
Browse files
app.py
CHANGED
|
@@ -1,9 +1,19 @@
|
|
| 1 |
-
from datasets import load_dataset
|
| 2 |
import streamlit as st
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
|
| 4 |
HF_API_TOKEN = st.secrets["HF_API_TOKEN"]
|
| 5 |
PROMPT_COLOR = "#CA437E"
|
| 6 |
|
|
|
|
| 7 |
def safe_text(text):
|
| 8 |
text = text.replace("\n", "<br>")
|
| 9 |
return f"<pre>{text}</pre>"
|
|
@@ -16,20 +26,44 @@ def prompt_markup_format(text):
|
|
| 16 |
def generation_markup_format(text):
|
| 17 |
return f"<font color={PROMPT_COLOR}>{text}</pre></font>"
|
| 18 |
|
| 19 |
-
|
|
|
|
| 20 |
ds = ds["train"]
|
| 21 |
|
| 22 |
possible_prompts = ds.unique("prompt")
|
| 23 |
-
chosen_prompt = st.selectbox("Chose a prompt", possible_prompts)
|
| 24 |
-
st.markdown(safe_text(chosen_prompt), unsafe_allow_html=True)
|
| 25 |
|
| 26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
|
|
|
|
|
|
|
|
|
|
| 28 |
|
| 29 |
-
index_sample = st.number_input("Index of the chosen example", min_value=0, max_value=len(sub_ds) - 1, value=0, step=1)
|
| 30 |
-
sample = sub_ds[index_sample]
|
| 31 |
-
markdown_text = generation_markup_format(safe_text(sample['generation']))
|
| 32 |
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import streamlit as st
|
| 2 |
+
from datasets import load_dataset
|
| 3 |
+
|
| 4 |
+
st.set_page_config(
|
| 5 |
+
page_icon="🧊",
|
| 6 |
+
layout="wide",
|
| 7 |
+
)
|
| 8 |
+
|
| 9 |
+
st.write(
|
| 10 |
+
"This is an application for viewing different generations for the same prompt. The generations vary depending on the checkpoint used and also the parameters used for the generation."
|
| 11 |
+
)
|
| 12 |
|
| 13 |
HF_API_TOKEN = st.secrets["HF_API_TOKEN"]
|
| 14 |
PROMPT_COLOR = "#CA437E"
|
| 15 |
|
| 16 |
+
|
| 17 |
def safe_text(text):
|
| 18 |
text = text.replace("\n", "<br>")
|
| 19 |
return f"<pre>{text}</pre>"
|
|
|
|
| 26 |
def generation_markup_format(text):
|
| 27 |
return f"<font color={PROMPT_COLOR}>{text}</pre></font>"
|
| 28 |
|
| 29 |
+
|
| 30 |
+
ds = load_dataset("SaulLu/bloom-generations", use_auth_token=HF_API_TOKEN)
|
| 31 |
ds = ds["train"]
|
| 32 |
|
| 33 |
possible_prompts = ds.unique("prompt")
|
|
|
|
|
|
|
| 34 |
|
| 35 |
+
col_1, col_2 = st.columns(2)
|
| 36 |
+
with col_1:
|
| 37 |
+
st.markdown("<h1 style='text-align: center'>Prompt</h1>", unsafe_allow_html=True)
|
| 38 |
+
chosen_prompt = st.selectbox("Chose a prompt", possible_prompts)
|
| 39 |
+
st.markdown(safe_text(chosen_prompt), unsafe_allow_html=True)
|
| 40 |
|
| 41 |
+
sub_ds = ds.filter(
|
| 42 |
+
lambda exs: [prompt == chosen_prompt for prompt in exs["prompt"]], batched=True
|
| 43 |
+
)
|
| 44 |
|
|
|
|
|
|
|
|
|
|
| 45 |
|
| 46 |
+
with col_2:
|
| 47 |
+
st.markdown(
|
| 48 |
+
"<h1 style='text-align: center'>Generation</h1>", unsafe_allow_html=True
|
| 49 |
+
)
|
| 50 |
+
index_sample = st.number_input(
|
| 51 |
+
"Index of the chosen example",
|
| 52 |
+
min_value=0,
|
| 53 |
+
max_value=len(sub_ds) - 1,
|
| 54 |
+
value=0,
|
| 55 |
+
step=1,
|
| 56 |
+
)
|
| 57 |
+
sample = sub_ds[index_sample]
|
| 58 |
+
markdown_text = generation_markup_format(safe_text(sample["generation"]))
|
| 59 |
+
st.markdown(markdown_text, unsafe_allow_html=True)
|
| 60 |
+
st.markdown(
|
| 61 |
+
"<h2 style='text-align: center'>Generation configuration</h2>",
|
| 62 |
+
unsafe_allow_html=True,
|
| 63 |
+
)
|
| 64 |
+
config = {
|
| 65 |
+
key: value
|
| 66 |
+
for key, value in sample.items()
|
| 67 |
+
if key not in ["prompt", "generation"]
|
| 68 |
+
}
|
| 69 |
+
config
|