reformat code
Browse files
app.py
CHANGED
|
@@ -8,6 +8,7 @@ import streamlit as st
|
|
| 8 |
MODELS = ["CodeParrot", "InCoder", "CodeGen", "PolyCoder"]
|
| 9 |
GENERATION_MODELS = ["CodeParrot", "InCoder", "CodeGen"]
|
| 10 |
|
|
|
|
| 11 |
@st.cache()
|
| 12 |
def load_examples():
|
| 13 |
with open("utils/examples.json", "r") as f:
|
|
@@ -18,10 +19,12 @@ def load_examples():
|
|
| 18 |
def read_markdown(path):
|
| 19 |
with open(path, "r") as f:
|
| 20 |
output = f.read()
|
| 21 |
-
st.markdown(output, unsafe_allow_html=True)
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
def generate_code(
|
|
|
|
|
|
|
| 25 |
# call space using its API endpoint
|
| 26 |
url = (
|
| 27 |
f"https://hf.space/embed/loubnabnl/{model_name.lower()}-subspace/+/api/predict/"
|
|
@@ -30,21 +33,33 @@ def generate_code(generations, model_name, gen_prompt, max_new_tokens, temperatu
|
|
| 30 |
url=url, json={"data": [gen_prompt, max_new_tokens, temperature, seed]}
|
| 31 |
)
|
| 32 |
generated_text = r.json()["data"][0]
|
| 33 |
-
generations.append(generated_text)
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
def generate_code_threads(
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
st.set_page_config(page_icon=":laptop:", layout="wide")
|
| 50 |
with open("utils/table_contents.txt", "r") as f:
|
|
@@ -59,12 +74,12 @@ read_markdown("utils/intro.txt")
|
|
| 59 |
st.subheader("1 - Code datasets")
|
| 60 |
read_markdown("datasets/intro.txt")
|
| 61 |
read_markdown("datasets/github_code.txt")
|
| 62 |
-
#GITHUB_CODE = "https://huggingface.co/datasets/lvwerra/github-code"
|
| 63 |
-
#st.markdown(f"Preview of some code files from Github repositories in [Github-code dataset]({GITHUB_CODE}):")
|
| 64 |
-
#df = pd.read_csv("utils/data_preview.csv")
|
| 65 |
-
#st.dataframe(df)
|
| 66 |
|
| 67 |
-
col1, col2= st.columns([1,2])
|
| 68 |
with col1:
|
| 69 |
selected_model = st.selectbox("", MODELS, key=1)
|
| 70 |
read_markdown(f"datasets/{selected_model.lower()}.txt")
|
|
@@ -73,7 +88,7 @@ read_markdown(f"datasets/{selected_model.lower()}.txt")
|
|
| 73 |
# Model architecture
|
| 74 |
st.subheader("2 - Model architecture")
|
| 75 |
read_markdown("architectures/intro.txt")
|
| 76 |
-
col1, col2= st.columns([1,2])
|
| 77 |
with col1:
|
| 78 |
selected_model = st.selectbox("", MODELS, key=2)
|
| 79 |
read_markdown(f"architectures/{selected_model.lower()}.txt")
|
|
@@ -85,12 +100,15 @@ read_markdown("evaluation/demo_humaneval.txt")
|
|
| 85 |
|
| 86 |
# Code generation
|
| 87 |
st.subheader("4 - Code generation ✨")
|
| 88 |
-
col1, col2, col3 = st.columns([7,1,6])
|
| 89 |
with col1:
|
| 90 |
st.markdown("**Models**")
|
| 91 |
selected_models = st.multiselect(
|
| 92 |
-
|
| 93 |
-
|
|
|
|
|
|
|
|
|
|
| 94 |
st.markdown(" ")
|
| 95 |
st.markdown("**Examples**")
|
| 96 |
examples = load_examples()
|
|
@@ -113,9 +131,7 @@ with col3:
|
|
| 113 |
step=4,
|
| 114 |
max_value=256,
|
| 115 |
)
|
| 116 |
-
seed = st.slider(
|
| 117 |
-
"Random seed:", value=42, min_value=0, step=1, max_value=1000
|
| 118 |
-
)
|
| 119 |
gen_prompt = st.text_area(
|
| 120 |
"Generate code with prompt:",
|
| 121 |
value=example_text,
|
|
@@ -141,4 +157,4 @@ if st.button("Generate code!"):
|
|
| 141 |
|
| 142 |
# Resources
|
| 143 |
st.subheader("Resources")
|
| 144 |
-
read_markdown("utils/resources.txt")
|
|
|
|
| 8 |
MODELS = ["CodeParrot", "InCoder", "CodeGen", "PolyCoder"]
|
| 9 |
GENERATION_MODELS = ["CodeParrot", "InCoder", "CodeGen"]
|
| 10 |
|
| 11 |
+
|
| 12 |
@st.cache()
|
| 13 |
def load_examples():
|
| 14 |
with open("utils/examples.json", "r") as f:
|
|
|
|
| 19 |
def read_markdown(path):
|
| 20 |
with open(path, "r") as f:
|
| 21 |
output = f.read()
|
| 22 |
+
st.markdown(output, unsafe_allow_html=True)
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
def generate_code(
|
| 26 |
+
generations, model_name, gen_prompt, max_new_tokens, temperature, seed
|
| 27 |
+
):
|
| 28 |
# call space using its API endpoint
|
| 29 |
url = (
|
| 30 |
f"https://hf.space/embed/loubnabnl/{model_name.lower()}-subspace/+/api/predict/"
|
|
|
|
| 33 |
url=url, json={"data": [gen_prompt, max_new_tokens, temperature, seed]}
|
| 34 |
)
|
| 35 |
generated_text = r.json()["data"][0]
|
| 36 |
+
generations.append(generated_text)
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
def generate_code_threads(
|
| 40 |
+
generations, models, gen_prompt, max_new_tokens, temperature, seed
|
| 41 |
+
):
|
| 42 |
+
threads = []
|
| 43 |
+
for model_name in models:
|
| 44 |
+
# create the thread
|
| 45 |
+
threads.append(
|
| 46 |
+
threading.Thread(
|
| 47 |
+
target=generate_code,
|
| 48 |
+
args=(
|
| 49 |
+
generations,
|
| 50 |
+
model_name,
|
| 51 |
+
gen_prompt,
|
| 52 |
+
max_new_tokens,
|
| 53 |
+
temperature,
|
| 54 |
+
seed,
|
| 55 |
+
),
|
| 56 |
+
)
|
| 57 |
+
)
|
| 58 |
+
threads[-1].start()
|
| 59 |
+
|
| 60 |
+
for t in threads:
|
| 61 |
+
t.join()
|
| 62 |
+
|
| 63 |
|
| 64 |
st.set_page_config(page_icon=":laptop:", layout="wide")
|
| 65 |
with open("utils/table_contents.txt", "r") as f:
|
|
|
|
| 74 |
st.subheader("1 - Code datasets")
|
| 75 |
read_markdown("datasets/intro.txt")
|
| 76 |
read_markdown("datasets/github_code.txt")
|
| 77 |
+
# GITHUB_CODE = "https://huggingface.co/datasets/lvwerra/github-code"
|
| 78 |
+
# st.markdown(f"Preview of some code files from Github repositories in [Github-code dataset]({GITHUB_CODE}):")
|
| 79 |
+
# df = pd.read_csv("utils/data_preview.csv")
|
| 80 |
+
# st.dataframe(df)
|
| 81 |
|
| 82 |
+
col1, col2 = st.columns([1, 2])
|
| 83 |
with col1:
|
| 84 |
selected_model = st.selectbox("", MODELS, key=1)
|
| 85 |
read_markdown(f"datasets/{selected_model.lower()}.txt")
|
|
|
|
| 88 |
# Model architecture
|
| 89 |
st.subheader("2 - Model architecture")
|
| 90 |
read_markdown("architectures/intro.txt")
|
| 91 |
+
col1, col2 = st.columns([1, 2])
|
| 92 |
with col1:
|
| 93 |
selected_model = st.selectbox("", MODELS, key=2)
|
| 94 |
read_markdown(f"architectures/{selected_model.lower()}.txt")
|
|
|
|
| 100 |
|
| 101 |
# Code generation
|
| 102 |
st.subheader("4 - Code generation ✨")
|
| 103 |
+
col1, col2, col3 = st.columns([7, 1, 6])
|
| 104 |
with col1:
|
| 105 |
st.markdown("**Models**")
|
| 106 |
selected_models = st.multiselect(
|
| 107 |
+
"Select code generation models to compare:",
|
| 108 |
+
GENERATION_MODELS,
|
| 109 |
+
default=["CodeParrot"],
|
| 110 |
+
key=3,
|
| 111 |
+
)
|
| 112 |
st.markdown(" ")
|
| 113 |
st.markdown("**Examples**")
|
| 114 |
examples = load_examples()
|
|
|
|
| 131 |
step=4,
|
| 132 |
max_value=256,
|
| 133 |
)
|
| 134 |
+
seed = st.slider("Random seed:", value=42, min_value=0, step=1, max_value=1000)
|
|
|
|
|
|
|
| 135 |
gen_prompt = st.text_area(
|
| 136 |
"Generate code with prompt:",
|
| 137 |
value=example_text,
|
|
|
|
| 157 |
|
| 158 |
# Resources
|
| 159 |
st.subheader("Resources")
|
| 160 |
+
read_markdown("utils/resources.txt")
|