Spaces:
Build error
Build error
Coyoteranger
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -1,115 +1,115 @@
|
|
1 |
-
import streamlit as st
|
2 |
-
from transformers import AutoModelForCausalLM, AutoTokenizer
|
3 |
-
|
4 |
-
# Load pre-trained model and tokenizer from the checkpoint
|
5 |
-
model_name = "
|
6 |
-
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
7 |
-
|
8 |
-
model = AutoModelForCausalLM.from_pretrained(model_name)
|
9 |
-
|
10 |
-
# Function to clean up repetitive lines in code
|
11 |
-
def clean_code_response(response):
|
12 |
-
lines = response.splitlines()
|
13 |
-
unique_lines = []
|
14 |
-
for line in lines:
|
15 |
-
if line.strip() not in unique_lines: # Avoid duplicates
|
16 |
-
unique_lines.append(line.strip())
|
17 |
-
return "\n".join(unique_lines)
|
18 |
-
|
19 |
-
# Function to generate Flutter code
|
20 |
-
def generate_flutter_code(prompt, temperature, top_p, max_length, num_return_sequences, repetition_penalty, top_k):
|
21 |
-
inputs = tokenizer(
|
22 |
-
prompt,
|
23 |
-
return_tensors="pt",
|
24 |
-
padding=True,
|
25 |
-
truncation=True,
|
26 |
-
)
|
27 |
-
outputs = model.generate(
|
28 |
-
inputs["input_ids"],
|
29 |
-
max_length=max_length,
|
30 |
-
num_return_sequences=num_return_sequences,
|
31 |
-
temperature=temperature,
|
32 |
-
top_p=top_p,
|
33 |
-
top_k=top_k,
|
34 |
-
repetition_penalty=repetition_penalty,
|
35 |
-
do_sample=True,
|
36 |
-
pad_token_id=tokenizer.pad_token_id,
|
37 |
-
)
|
38 |
-
code = [tokenizer.decode(output, skip_special_tokens=True) for output in outputs]
|
39 |
-
return [clean_code_response(c) for c in code]
|
40 |
-
|
41 |
-
# App Title
|
42 |
-
st.title("Flutter Code Generator")
|
43 |
-
|
44 |
-
# Default parameter values
|
45 |
-
DEFAULT_TEMPERATURE = 0.7
|
46 |
-
DEFAULT_TOP_P = 0.9
|
47 |
-
DEFAULT_MAX_LENGTH = 512
|
48 |
-
DEFAULT_NUM_RETURN_SEQUENCES = 1
|
49 |
-
DEFAULT_REPETITION_PENALTY = 1.2
|
50 |
-
DEFAULT_TOP_K = 50
|
51 |
-
|
52 |
-
# Sidebar for settings
|
53 |
-
st.sidebar.title("Generation Settings")
|
54 |
-
|
55 |
-
temperature = st.sidebar.slider(
|
56 |
-
"Temperature (randomness)",
|
57 |
-
0.1, 1.0, DEFAULT_TEMPERATURE, step=0.1,
|
58 |
-
)
|
59 |
-
|
60 |
-
top_p = st.sidebar.slider(
|
61 |
-
"Top-p (cumulative probability)",
|
62 |
-
0.1, 1.0, DEFAULT_TOP_P, step=0.1,
|
63 |
-
)
|
64 |
-
|
65 |
-
max_length = st.sidebar.slider(
|
66 |
-
"Max Output Length (tokens)",
|
67 |
-
128, 1024, DEFAULT_MAX_LENGTH, step=64,
|
68 |
-
)
|
69 |
-
|
70 |
-
num_return_sequences = st.sidebar.slider(
|
71 |
-
"Number of Outputs",
|
72 |
-
1, 5, DEFAULT_NUM_RETURN_SEQUENCES,
|
73 |
-
)
|
74 |
-
|
75 |
-
repetition_penalty = st.sidebar.slider(
|
76 |
-
"Repetition Penalty",
|
77 |
-
1.0, 2.0, DEFAULT_REPETITION_PENALTY, step=0.1,
|
78 |
-
)
|
79 |
-
|
80 |
-
top_k = st.sidebar.slider(
|
81 |
-
"Top-k (limit sampling pool)",
|
82 |
-
0, 100, DEFAULT_TOP_K,
|
83 |
-
)
|
84 |
-
|
85 |
-
# Reset to defaults button
|
86 |
-
if st.sidebar.button("Reset to Defaults"):
|
87 |
-
st.session_state.update(
|
88 |
-
{
|
89 |
-
"temperature": DEFAULT_TEMPERATURE,
|
90 |
-
"top_p": DEFAULT_TOP_P,
|
91 |
-
"max_length": DEFAULT_MAX_LENGTH,
|
92 |
-
"num_return_sequences": DEFAULT_NUM_RETURN_SEQUENCES,
|
93 |
-
"repetition_penalty": DEFAULT_REPETITION_PENALTY,
|
94 |
-
"top_k": DEFAULT_TOP_K,
|
95 |
-
}
|
96 |
-
)
|
97 |
-
|
98 |
-
# Input Section
|
99 |
-
user_input = st.text_area(
|
100 |
-
"Enter your prompt (e.g., 'Create a responsive login screen'):",
|
101 |
-
max_chars=200,
|
102 |
-
)
|
103 |
-
|
104 |
-
# Output Section
|
105 |
-
if st.button("Generate Code"):
|
106 |
-
if user_input.strip():
|
107 |
-
prompt = f"{user_input.strip()}"
|
108 |
-
generated_code = generate_flutter_code(
|
109 |
-
prompt, temperature, top_p, max_length, num_return_sequences, repetition_penalty, top_k
|
110 |
-
)
|
111 |
-
for i, code in enumerate(generated_code, start=1):
|
112 |
-
st.subheader(f"Output {i}")
|
113 |
-
st.code(code, language="dart")
|
114 |
-
else:
|
115 |
-
st.error("Please enter a prompt before clicking 'Generate Code'.")
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
3 |
+
|
4 |
+
# Load pre-trained model and tokenizer from the checkpoint
|
5 |
+
model_name = "flutter-code-generator/flutter_codegen_model/checkpoint-1500"
|
6 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
7 |
+
model = AutoModelForCausalLM.from_pretrained(model_name)
|
8 |
+
model = AutoModelForCausalLM.from_pretrained(model_name)
|
9 |
+
|
10 |
+
# Function to clean up repetitive lines in code
|
11 |
+
def clean_code_response(response):
|
12 |
+
lines = response.splitlines()
|
13 |
+
unique_lines = []
|
14 |
+
for line in lines:
|
15 |
+
if line.strip() not in unique_lines: # Avoid duplicates
|
16 |
+
unique_lines.append(line.strip())
|
17 |
+
return "\n".join(unique_lines)
|
18 |
+
|
19 |
+
# Function to generate Flutter code
|
20 |
+
def generate_flutter_code(prompt, temperature, top_p, max_length, num_return_sequences, repetition_penalty, top_k):
|
21 |
+
inputs = tokenizer(
|
22 |
+
prompt,
|
23 |
+
return_tensors="pt",
|
24 |
+
padding=True,
|
25 |
+
truncation=True,
|
26 |
+
)
|
27 |
+
outputs = model.generate(
|
28 |
+
inputs["input_ids"],
|
29 |
+
max_length=max_length,
|
30 |
+
num_return_sequences=num_return_sequences,
|
31 |
+
temperature=temperature,
|
32 |
+
top_p=top_p,
|
33 |
+
top_k=top_k,
|
34 |
+
repetition_penalty=repetition_penalty,
|
35 |
+
do_sample=True,
|
36 |
+
pad_token_id=tokenizer.pad_token_id,
|
37 |
+
)
|
38 |
+
code = [tokenizer.decode(output, skip_special_tokens=True) for output in outputs]
|
39 |
+
return [clean_code_response(c) for c in code]
|
40 |
+
|
41 |
+
# App Title
|
42 |
+
st.title("Flutter Code Generator")
|
43 |
+
|
44 |
+
# Default parameter values
|
45 |
+
DEFAULT_TEMPERATURE = 0.7
|
46 |
+
DEFAULT_TOP_P = 0.9
|
47 |
+
DEFAULT_MAX_LENGTH = 512
|
48 |
+
DEFAULT_NUM_RETURN_SEQUENCES = 1
|
49 |
+
DEFAULT_REPETITION_PENALTY = 1.2
|
50 |
+
DEFAULT_TOP_K = 50
|
51 |
+
|
52 |
+
# Sidebar for settings
|
53 |
+
st.sidebar.title("Generation Settings")
|
54 |
+
|
55 |
+
temperature = st.sidebar.slider(
|
56 |
+
"Temperature (randomness)",
|
57 |
+
0.1, 1.0, DEFAULT_TEMPERATURE, step=0.1,
|
58 |
+
)
|
59 |
+
|
60 |
+
top_p = st.sidebar.slider(
|
61 |
+
"Top-p (cumulative probability)",
|
62 |
+
0.1, 1.0, DEFAULT_TOP_P, step=0.1,
|
63 |
+
)
|
64 |
+
|
65 |
+
max_length = st.sidebar.slider(
|
66 |
+
"Max Output Length (tokens)",
|
67 |
+
128, 1024, DEFAULT_MAX_LENGTH, step=64,
|
68 |
+
)
|
69 |
+
|
70 |
+
num_return_sequences = st.sidebar.slider(
|
71 |
+
"Number of Outputs",
|
72 |
+
1, 5, DEFAULT_NUM_RETURN_SEQUENCES,
|
73 |
+
)
|
74 |
+
|
75 |
+
repetition_penalty = st.sidebar.slider(
|
76 |
+
"Repetition Penalty",
|
77 |
+
1.0, 2.0, DEFAULT_REPETITION_PENALTY, step=0.1,
|
78 |
+
)
|
79 |
+
|
80 |
+
top_k = st.sidebar.slider(
|
81 |
+
"Top-k (limit sampling pool)",
|
82 |
+
0, 100, DEFAULT_TOP_K,
|
83 |
+
)
|
84 |
+
|
85 |
+
# Reset to defaults button
|
86 |
+
if st.sidebar.button("Reset to Defaults"):
|
87 |
+
st.session_state.update(
|
88 |
+
{
|
89 |
+
"temperature": DEFAULT_TEMPERATURE,
|
90 |
+
"top_p": DEFAULT_TOP_P,
|
91 |
+
"max_length": DEFAULT_MAX_LENGTH,
|
92 |
+
"num_return_sequences": DEFAULT_NUM_RETURN_SEQUENCES,
|
93 |
+
"repetition_penalty": DEFAULT_REPETITION_PENALTY,
|
94 |
+
"top_k": DEFAULT_TOP_K,
|
95 |
+
}
|
96 |
+
)
|
97 |
+
|
98 |
+
# Input Section
|
99 |
+
user_input = st.text_area(
|
100 |
+
"Enter your prompt (e.g., 'Create a responsive login screen'):",
|
101 |
+
max_chars=200,
|
102 |
+
)
|
103 |
+
|
104 |
+
# Output Section
|
105 |
+
if st.button("Generate Code"):
|
106 |
+
if user_input.strip():
|
107 |
+
prompt = f"{user_input.strip()}"
|
108 |
+
generated_code = generate_flutter_code(
|
109 |
+
prompt, temperature, top_p, max_length, num_return_sequences, repetition_penalty, top_k
|
110 |
+
)
|
111 |
+
for i, code in enumerate(generated_code, start=1):
|
112 |
+
st.subheader(f"Output {i}")
|
113 |
+
st.code(code, language="dart")
|
114 |
+
else:
|
115 |
+
st.error("Please enter a prompt before clicking 'Generate Code'.")
|