Spaces:
Runtime error
Runtime error
shrimantasatpati
commited on
Commit
•
0b252a0
1
Parent(s):
3b4796e
Updated app.py
Browse files
app.py
CHANGED
@@ -2,6 +2,12 @@ import streamlit as st
|
|
2 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
3 |
import torch
|
4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
# Load the Phi 2 model and tokenizer
|
6 |
tokenizer = AutoTokenizer.from_pretrained(
|
7 |
"microsoft/phi-2",
|
@@ -10,26 +16,28 @@ tokenizer = AutoTokenizer.from_pretrained(
|
|
10 |
|
11 |
model = AutoModelForCausalLM.from_pretrained(
|
12 |
"microsoft/phi-2",
|
13 |
-
device_map="
|
14 |
trust_remote_code=True,
|
15 |
-
offload_folder="offload"
|
|
|
16 |
)
|
17 |
|
18 |
# Streamlit UI
|
19 |
st.title("Microsoft Phi 2 Streamlit App")
|
20 |
|
21 |
# User input prompt
|
22 |
-
prompt = st.text_area("Enter your prompt:", """Write a
|
23 |
|
24 |
# Generate output based on user input
|
25 |
if st.button("Generate Output"):
|
26 |
with torch.no_grad():
|
27 |
-
token_ids = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt")
|
28 |
output_ids = model.generate(
|
29 |
token_ids.to(model.device),
|
30 |
max_new_tokens=512,
|
31 |
do_sample=True,
|
32 |
-
temperature=0.3
|
|
|
33 |
)
|
34 |
|
35 |
output = tokenizer.decode(output_ids[0][token_ids.size(1):])
|
|
|
2 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
3 |
import torch
|
4 |
|
5 |
+
import torch
|
6 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
7 |
+
import gradio as gr
|
8 |
+
# torch.set_default_device("cuda")
|
9 |
+
|
10 |
+
|
11 |
# Load the Phi 2 model and tokenizer
|
12 |
tokenizer = AutoTokenizer.from_pretrained(
|
13 |
"microsoft/phi-2",
|
|
|
16 |
|
17 |
model = AutoModelForCausalLM.from_pretrained(
|
18 |
"microsoft/phi-2",
|
19 |
+
device_map="cpu",
|
20 |
trust_remote_code=True,
|
21 |
+
# offload_folder="offload",
|
22 |
+
torch_dtype=torch.float32
|
23 |
)
|
24 |
|
25 |
# Streamlit UI
|
26 |
st.title("Microsoft Phi 2 Streamlit App")
|
27 |
|
28 |
# User input prompt
|
29 |
+
prompt = st.text_area("Enter your prompt:", """Write a short summary about how to create a healthy lifestyle.""")
|
30 |
|
31 |
# Generate output based on user input
|
32 |
if st.button("Generate Output"):
|
33 |
with torch.no_grad():
|
34 |
+
token_ids = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt", return_attention_mask=False)
|
35 |
output_ids = model.generate(
|
36 |
token_ids.to(model.device),
|
37 |
max_new_tokens=512,
|
38 |
do_sample=True,
|
39 |
+
temperature=0.3,
|
40 |
+
max_length=200
|
41 |
)
|
42 |
|
43 |
output = tokenizer.decode(output_ids[0][token_ids.size(1):])
|