Spaces:
Running
Running
Update src/streamlit_app.py
Browse files- src/streamlit_app.py +6 -6
src/streamlit_app.py
CHANGED
@@ -10,11 +10,16 @@ st.title("Gemini Token Probabilities")
|
|
10 |
# --- Load the SentencePiece tokenizer once ---
|
11 |
@st.cache_resource
|
12 |
def load_tokenizer():
|
13 |
-
|
|
|
|
|
|
|
14 |
model_path = os.path.join(model_dir, "gemini-1.5-pro-002.spm.model")
|
|
|
15 |
if not os.path.isfile(model_path):
|
16 |
st.error(f"Cannot find model at:\n{model_path}")
|
17 |
st.stop()
|
|
|
18 |
sp = spm.SentencePieceProcessor()
|
19 |
sp.Load(model_path)
|
20 |
return sp
|
@@ -57,11 +62,6 @@ if st.button("Tokenize"):
|
|
57 |
})
|
58 |
|
59 |
# 5) Display with progress bars (as percentages)
|
60 |
-
#
|
61 |
-
# - format="percent" tells Streamlit to multiply the 0β1 value by 100 internally
|
62 |
-
# and append β%β to the label.
|
63 |
-
# - min_value=0 and max_value=1 anchor the range.
|
64 |
-
#st.markdown("### Tokenization Result (Global 0β100 %)")
|
65 |
st.dataframe(
|
66 |
df,
|
67 |
use_container_width=True,
|
|
|
10 |
# --- Load the SentencePiece tokenizer once ---
|
11 |
@st.cache_resource
|
12 |
def load_tokenizer():
|
13 |
+
# Determine the directory that this script lives in (i.e. src/)
|
14 |
+
here = os.path.dirname(__file__)
|
15 |
+
# Build the absolute path to the gemini-1.5-pro-002 folder inside src/
|
16 |
+
model_dir = os.path.join(here, "gemini-1.5-pro-002")
|
17 |
model_path = os.path.join(model_dir, "gemini-1.5-pro-002.spm.model")
|
18 |
+
|
19 |
if not os.path.isfile(model_path):
|
20 |
st.error(f"Cannot find model at:\n{model_path}")
|
21 |
st.stop()
|
22 |
+
|
23 |
sp = spm.SentencePieceProcessor()
|
24 |
sp.Load(model_path)
|
25 |
return sp
|
|
|
62 |
})
|
63 |
|
64 |
# 5) Display with progress bars (as percentages)
|
|
|
|
|
|
|
|
|
|
|
65 |
st.dataframe(
|
66 |
df,
|
67 |
use_container_width=True,
|