File size: 1,796 Bytes
826e275 7a75a15 826e275 d73a8e9 826e275 6e4f775 7a75a15 826e275 e0bb50d 49efed6 b598d9f 826e275 6e4f775 49efed6 6e4f775 943d773 826e275 49efed6 826e275 49efed6 826e275 49efed6 0e07a66 49efed6 0e07a66 d73a8e9 826e275 49efed6 b1dd47e 826e275 b1dd47e 0499581 d73a8e9 b1dd47e 49efed6 9c8dd72 |
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 |
"""
The Streamlit app for the project demo.
In the demo, the user can write a prompt
and the model will generate a response using the grouped sampling algorithm.
"""
import streamlit as st
from torch.cuda import CudaError
from available_models import AVAILABLE_MODELS
from hanlde_form_submit import on_form_submit
st.title("讚讙讬诪讛 讘拽讘讜爪讜转 - 砖讬诪讜砖 讬注讬诇 讘诪讜讚诇讬 砖驻讛 住讬讘转讬讬诐")
with st.form("request_form"):
selected_model_name: str = st.selectbox(
label="讘讞专讜 诪讜讚诇",
options=AVAILABLE_MODELS,
help="llama-30b-hf generates better texts but is slower",
)
output_length: int = st.number_input(
label="讻诪讜转 讛诪讬诇讬诐 讛诪拽住讬诪诇讬转 讘驻诇讟 - 讘讬谉 1 诇-4096",
min_value=1,
max_value=4096,
value=5,
)
submitted_prompt: str = st.text_area(
label="讛拽诇讟 诇讗诇讜讙专讬转诐 (讘讗谞讙诇讬转 讘诇讘讚)",
value="Instruction: Answer in yes or no.\n"
"Question: Is the sky blue?\n"
"Answer: ",
max_chars=2048,
)
submitted: bool = st.form_submit_button(
label="爪讜专 讟拽住讟",
disabled=False,
)
if submitted:
try:
output = on_form_submit(
selected_model_name,
output_length,
submitted_prompt,
)
except CudaError as e:
st.error("Out of memory. Please try a smaller model, shorter prompt, or a smaller output length.")
except (ValueError, TypeError, RuntimeError) as e:
st.error(e)
else:
st.write(f"Generated text: {output}")
with open("user_instructions_hebrew.md", "r") as fh:
long_description = fh.read()
st.markdown(long_description)
|