make blog
Browse files
app.py
CHANGED
@@ -53,4 +53,79 @@ if selected_task == " ":
|
|
53 |
st.title("Code Generation Models")
|
54 |
with open("utils/intro.txt", "r") as f:
|
55 |
intro = f.read()
|
56 |
-
st.markdown(intro)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
53 |
st.title("Code Generation Models")
|
54 |
with open("utils/intro.txt", "r") as f:
|
55 |
intro = f.read()
|
56 |
+
st.markdown(intro)
|
57 |
+
|
58 |
+
elif selected_task == "Pretraining datasets":
|
59 |
+
st.title("Pretraining datasets π")
|
60 |
+
st.markdown(
|
61 |
+
f"Preview of some code files from Github repositories in [Github-code dataset]({GITHUB_CODE}):"
|
62 |
+
)
|
63 |
+
df = pd.read_csv("utils/data_preview.csv")
|
64 |
+
st.dataframe(df)
|
65 |
+
for model in selected_models:
|
66 |
+
with open(f"datasets/{model.lower()}.txt", "r") as f:
|
67 |
+
text = f.read()
|
68 |
+
st.markdown(f"### {model}")
|
69 |
+
st.markdown(text)
|
70 |
+
|
71 |
+
elif selected_task == "Model architecture":
|
72 |
+
st.title("Model architecture")
|
73 |
+
for model in selected_models:
|
74 |
+
with open(f"architectures/{model.lower()}.txt", "r") as f:
|
75 |
+
text = f.read()
|
76 |
+
st.markdown(f"## {model}")
|
77 |
+
st.markdown(text)
|
78 |
+
if model == "InCoder":
|
79 |
+
st.image(INCODER_IMG, caption="Figure 1: InCoder training", width=700)
|
80 |
+
|
81 |
+
elif selected_task == "Model evaluation":
|
82 |
+
st.title("Code models evaluation π")
|
83 |
+
with open("evaluation/intro.txt", "r") as f:
|
84 |
+
intro = f.read()
|
85 |
+
st.markdown(intro)
|
86 |
+
|
87 |
+
elif selected_task == "Code generation":
|
88 |
+
st.title("Code generation π»")
|
89 |
+
st.sidebar.header("Examples")
|
90 |
+
examples = load_examples()
|
91 |
+
example_names = [example["name"] for example in examples]
|
92 |
+
name2id = dict([(name, i) for i, name in enumerate(example_names)])
|
93 |
+
selected_example = st.sidebar.selectbox(
|
94 |
+
"Select one of the following examples or implement yours", example_names
|
95 |
+
)
|
96 |
+
example_text = examples[name2id[selected_example]]["value"]
|
97 |
+
default_length = examples[name2id[selected_example]]["length"]
|
98 |
+
st.sidebar.header("Generation settings")
|
99 |
+
temperature = st.sidebar.slider(
|
100 |
+
"Temperature:", value=0.2, min_value=0.0, step=0.1, max_value=2.0
|
101 |
+
)
|
102 |
+
max_new_tokens = st.sidebar.slider(
|
103 |
+
"Number of tokens to generate:",
|
104 |
+
value=default_length,
|
105 |
+
min_value=8,
|
106 |
+
step=8,
|
107 |
+
max_value=256,
|
108 |
+
)
|
109 |
+
seed = st.sidebar.slider(
|
110 |
+
"Random seed:", value=42, min_value=0, step=1, max_value=1000
|
111 |
+
)
|
112 |
+
gen_prompt = st.text_area(
|
113 |
+
"Generate code with prompt:",
|
114 |
+
value=example_text,
|
115 |
+
height=220,
|
116 |
+
).strip()
|
117 |
+
if st.button("Generate code!"):
|
118 |
+
with st.spinner("Generating code..."):
|
119 |
+
# Create a multiprocessing Pool
|
120 |
+
pool = Pool()
|
121 |
+
generate_parallel = partial(
|
122 |
+
generate_code,
|
123 |
+
gen_prompt=gen_prompt,
|
124 |
+
max_new_tokens=max_new_tokens,
|
125 |
+
temperature=temperature,
|
126 |
+
seed=seed,
|
127 |
+
)
|
128 |
+
output = pool.map(generate_parallel, selected_models)
|
129 |
+
for i in range(len(output)):
|
130 |
+
st.markdown(f"**{selected_models[i]}**")
|
131 |
+
st.code(output[i])
|