Spaces:
Runtime error
Runtime error
Xianbao QIAN
commited on
Commit
·
93e4eee
1
Parent(s):
d76943e
update new ui
Browse files
app.py
CHANGED
@@ -7,41 +7,102 @@ import sambanova
|
|
7 |
def generate(
|
8 |
message: str,
|
9 |
chat_history: list[tuple[str, str]],
|
10 |
-
|
|
|
11 |
temperature: float = 0.6,
|
12 |
top_p: float = 0.9,
|
13 |
top_k: int = 50,
|
14 |
repetition_penalty: float = 1.2,
|
15 |
) -> Iterator[str]:
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
conversation.append({"role": "user", "content": message})
|
25 |
|
26 |
outputs = []
|
27 |
-
for text in sambanova.Streamer(conversation,
|
28 |
-
|
|
|
|
|
|
|
29 |
outputs.append(text)
|
30 |
yield "".join(outputs)
|
31 |
|
32 |
-
|
33 |
-
|
|
|
34 |
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
|
35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
chat_interface = gr.ChatInterface(
|
37 |
-
|
38 |
additional_inputs=[
|
|
|
|
|
39 |
gr.Slider(
|
40 |
-
label="Max
|
41 |
minimum=1,
|
42 |
-
maximum=
|
43 |
step=1,
|
44 |
-
value=
|
45 |
),
|
46 |
gr.Slider(
|
47 |
label="Temperature",
|
@@ -64,29 +125,26 @@ chat_interface = gr.ChatInterface(
|
|
64 |
step=1,
|
65 |
value=50,
|
66 |
),
|
67 |
-
|
68 |
-
label="Repetition penalty",
|
69 |
-
minimum=1.0,
|
70 |
-
maximum=2.0,
|
71 |
-
step=0.05,
|
72 |
-
value=1.2,
|
73 |
-
),
|
74 |
],
|
75 |
-
stop_btn=None,
|
76 |
-
fill_height=True,
|
77 |
examples=[
|
78 |
["Which one is bigger? 4.9 or 4.11"],
|
79 |
-
[
|
|
|
|
|
80 |
["Explain the plot of Cinderella in a sentence."],
|
81 |
["How many hours does it take a man to eat a Helicopter?"],
|
82 |
-
[
|
|
|
|
|
83 |
],
|
84 |
cache_examples=False,
|
85 |
)
|
|
|
86 |
with gr.Blocks() as demo:
|
87 |
gr.Markdown('# Sambanova model inference LLAMA 405B')
|
88 |
|
89 |
chat_interface.render()
|
90 |
-
|
91 |
if __name__ == "__main__":
|
92 |
demo.queue(max_size=20).launch()
|
|
|
7 |
def generate(
|
8 |
message: str,
|
9 |
chat_history: list[tuple[str, str]],
|
10 |
+
system_message,
|
11 |
+
max_tokens: int = 1024,
|
12 |
temperature: float = 0.6,
|
13 |
top_p: float = 0.9,
|
14 |
top_k: int = 50,
|
15 |
repetition_penalty: float = 1.2,
|
16 |
) -> Iterator[str]:
|
17 |
+
|
18 |
+
conversation = [{"role": "system", "content": system_message}]
|
19 |
+
|
20 |
+
for val in chat_history:
|
21 |
+
if val[0]:
|
22 |
+
conversation.append({"role": "user", "content": val[0]})
|
23 |
+
if val[1]:
|
24 |
+
conversation.append({"role": "assistant", "content": val[1]})
|
|
|
25 |
|
26 |
outputs = []
|
27 |
+
for text in sambanova.Streamer(conversation,
|
28 |
+
new_tokens=max_tokens,
|
29 |
+
temperature=temperature,
|
30 |
+
top_k=top_k,
|
31 |
+
top_p=top_p):
|
32 |
outputs.append(text)
|
33 |
yield "".join(outputs)
|
34 |
|
35 |
+
|
36 |
+
MAX_MAX_TOKENS = 2048
|
37 |
+
DEFAULT_MAX_TOKENS = 1024
|
38 |
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
|
39 |
|
40 |
+
# chat_interface = gr.ChatInterface(
|
41 |
+
# fn=generate,
|
42 |
+
# additional_inputs=[
|
43 |
+
# gr.Slider(
|
44 |
+
# label="Max new tokens",
|
45 |
+
# minimum=1,
|
46 |
+
# maximum=MAX_MAX_NEW_TOKENS,
|
47 |
+
# step=1,
|
48 |
+
# value=DEFAULT_MAX_NEW_TOKENS,
|
49 |
+
# ),
|
50 |
+
# gr.Slider(
|
51 |
+
# label="Temperature",
|
52 |
+
# minimum=0.1,
|
53 |
+
# maximum=4.0,
|
54 |
+
# step=0.1,
|
55 |
+
# value=0.6,
|
56 |
+
# ),
|
57 |
+
# gr.Slider(
|
58 |
+
# label="Top-p (nucleus sampling)",
|
59 |
+
# minimum=0.05,
|
60 |
+
# maximum=1.0,
|
61 |
+
# step=0.05,
|
62 |
+
# value=0.9,
|
63 |
+
# ),
|
64 |
+
# gr.Slider(
|
65 |
+
# label="Top-k",
|
66 |
+
# minimum=1,
|
67 |
+
# maximum=1000,
|
68 |
+
# step=1,
|
69 |
+
# value=50,
|
70 |
+
# ),
|
71 |
+
# gr.Slider(
|
72 |
+
# label="Repetition penalty",
|
73 |
+
# minimum=1.0,
|
74 |
+
# maximum=2.0,
|
75 |
+
# step=0.05,
|
76 |
+
# value=1.2,
|
77 |
+
# ),
|
78 |
+
# ],
|
79 |
+
# stop_btn=None,
|
80 |
+
# fill_height=True,
|
81 |
+
# examples=[
|
82 |
+
# ["Which one is bigger? 4.9 or 4.11"],
|
83 |
+
# [
|
84 |
+
# "Can you explain briefly to me what is the Python programming language?"
|
85 |
+
# ],
|
86 |
+
# ["Explain the plot of Cinderella in a sentence."],
|
87 |
+
# ["How many hours does it take a man to eat a Helicopter?"],
|
88 |
+
# [
|
89 |
+
# "Write a 100-word article on 'Benefits of Open-Source in AI research'"
|
90 |
+
# ],
|
91 |
+
# ],
|
92 |
+
# cache_examples=False,
|
93 |
+
# )
|
94 |
+
|
95 |
chat_interface = gr.ChatInterface(
|
96 |
+
generate,
|
97 |
additional_inputs=[
|
98 |
+
gr.Textbox(value="You are a friendly Chatbot.",
|
99 |
+
label="System message"),
|
100 |
gr.Slider(
|
101 |
+
label="Max tokens",
|
102 |
minimum=1,
|
103 |
+
maximum=MAX_MAX_TOKENS,
|
104 |
step=1,
|
105 |
+
value=DEFAULT_MAX_TOKENS,
|
106 |
),
|
107 |
gr.Slider(
|
108 |
label="Temperature",
|
|
|
125 |
step=1,
|
126 |
value=50,
|
127 |
),
|
128 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
129 |
],
|
|
|
|
|
130 |
examples=[
|
131 |
["Which one is bigger? 4.9 or 4.11"],
|
132 |
+
[
|
133 |
+
"Can you explain briefly to me what is the Python programming language?"
|
134 |
+
],
|
135 |
["Explain the plot of Cinderella in a sentence."],
|
136 |
["How many hours does it take a man to eat a Helicopter?"],
|
137 |
+
[
|
138 |
+
"Write a 100-word article on 'Benefits of Open-Source in AI research'"
|
139 |
+
],
|
140 |
],
|
141 |
cache_examples=False,
|
142 |
)
|
143 |
+
|
144 |
with gr.Blocks() as demo:
|
145 |
gr.Markdown('# Sambanova model inference LLAMA 405B')
|
146 |
|
147 |
chat_interface.render()
|
148 |
+
|
149 |
if __name__ == "__main__":
|
150 |
demo.queue(max_size=20).launch()
|