Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,11 +1,30 @@
|
|
1 |
import spaces
|
|
|
|
|
2 |
import gradio as gr
|
3 |
-
from huggingface_hub import
|
4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
"""
|
6 |
-
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
|
7 |
-
"""
|
8 |
-
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
|
9 |
|
10 |
@spaces.GPU(duration=120)
|
11 |
def respond(
|
@@ -15,39 +34,58 @@ def respond(
|
|
15 |
max_tokens,
|
16 |
temperature,
|
17 |
top_p,
|
|
|
18 |
):
|
19 |
-
|
|
|
|
|
|
|
|
|
|
|
20 |
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
|
27 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
-
|
30 |
|
31 |
-
for
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
|
|
39 |
|
40 |
-
|
41 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
|
43 |
-
"""
|
44 |
-
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
|
45 |
-
"""
|
46 |
demo = gr.ChatInterface(
|
47 |
respond,
|
48 |
additional_inputs=[
|
49 |
-
gr.
|
50 |
-
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
51 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
52 |
gr.Slider(
|
53 |
minimum=0.1,
|
@@ -56,9 +94,24 @@ demo = gr.ChatInterface(
|
|
56 |
step=0.05,
|
57 |
label="Top-p (nucleus sampling)",
|
58 |
),
|
|
|
59 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
)
|
61 |
|
62 |
-
|
63 |
if __name__ == "__main__":
|
64 |
-
demo.launch()
|
|
|
1 |
import spaces
|
2 |
+
import json
|
3 |
+
import subprocess
|
4 |
import gradio as gr
|
5 |
+
from huggingface_hub import hf_hub_download
|
6 |
|
7 |
+
subprocess.run('pip install llama-cpp-python==0.2.75 --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cu124', shell=True)
|
8 |
+
subprocess.run('pip install llama-cpp-agent==0.2.10', shell=True)
|
9 |
+
|
10 |
+
hf_hub_download(repo_id="bartowski/dolphin-2.9.1-yi-1.5-34b-GGUF", filename="dolphin-2.9.1-yi-1.5-34b-Q6_K.gguf", local_dir = "./models")
|
11 |
+
hf_hub_download(repo_id="bartowski/dolphin-2.9.1-yi-1.5-9b-GGUF", filename="dolphin-2.9.1-yi-1.5-9b-f32.gguf", local_dir = "./models")
|
12 |
+
|
13 |
+
css = """
|
14 |
+
.message-row {
|
15 |
+
justify-content: space-evenly !important;
|
16 |
+
}
|
17 |
+
.message-bubble-border {
|
18 |
+
border-radius: 6px !important;
|
19 |
+
border-color: #343140 !important;
|
20 |
+
}
|
21 |
+
.user {
|
22 |
+
background: #1e1c26 !important;
|
23 |
+
}
|
24 |
+
.assistant, .pending {
|
25 |
+
background: #16141c !important;
|
26 |
+
}
|
27 |
"""
|
|
|
|
|
|
|
28 |
|
29 |
@spaces.GPU(duration=120)
|
30 |
def respond(
|
|
|
34 |
max_tokens,
|
35 |
temperature,
|
36 |
top_p,
|
37 |
+
model,
|
38 |
):
|
39 |
+
from llama_cpp import Llama
|
40 |
+
from llama_cpp_agent import LlamaCppAgent
|
41 |
+
from llama_cpp_agent import MessagesFormatterType
|
42 |
+
from llama_cpp_agent.providers import LlamaCppPythonProvider
|
43 |
+
from llama_cpp_agent.chat_history import BasicChatHistory
|
44 |
+
from llama_cpp_agent.chat_history.messages import Roles
|
45 |
|
46 |
+
llm = Llama(
|
47 |
+
model_path=f"models/{model}",
|
48 |
+
n_gpu_layers=81,
|
49 |
+
)
|
50 |
+
provider = LlamaCppPythonProvider(llm)
|
51 |
|
52 |
+
agent = LlamaCppAgent(
|
53 |
+
provider,
|
54 |
+
system_prompt="You are a helpful assistant.",
|
55 |
+
predefined_messages_formatter_type=MessagesFormatterType.CHATML,
|
56 |
+
debug_output=True
|
57 |
+
)
|
58 |
+
|
59 |
+
settings = provider.get_provider_default_settings()
|
60 |
+
settings.max_tokens = max_tokens
|
61 |
+
settings.stream = True
|
62 |
|
63 |
+
messages = BasicChatHistory()
|
64 |
|
65 |
+
for msn in history:
|
66 |
+
user = {
|
67 |
+
'role': Roles.user,
|
68 |
+
'content': msn[0]
|
69 |
+
}
|
70 |
+
assistant = {
|
71 |
+
'role': Roles.assistant,
|
72 |
+
'content': msn[1]
|
73 |
+
}
|
74 |
|
75 |
+
messages.add_message(user)
|
76 |
+
messages.add_message(assistant)
|
77 |
+
|
78 |
+
stream = agent.get_chat_response(message, llm_sampling_settings=settings, chat_history=messages, returns_streaming_generator=True, print_output=False)
|
79 |
+
|
80 |
+
outputs = ""
|
81 |
+
for output in stream:
|
82 |
+
outputs += output
|
83 |
+
yield outputs
|
84 |
|
|
|
|
|
|
|
85 |
demo = gr.ChatInterface(
|
86 |
respond,
|
87 |
additional_inputs=[
|
88 |
+
gr.Slider(minimum=1, maximum=8192, value=8192, step=1, label="Max new tokens"),
|
|
|
89 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
90 |
gr.Slider(
|
91 |
minimum=0.1,
|
|
|
94 |
step=0.05,
|
95 |
label="Top-p (nucleus sampling)",
|
96 |
),
|
97 |
+
gr.Dropdown(['dolphin-2.9.1-yi-1.5-34b-Q6_K.gguf', 'dolphin-2.9.1-yi-1.5-9b-f32.gguf'], value="dolphin-2.9.1-yi-1.5-34b-Q6_K.gguf", label="Model"),
|
98 |
],
|
99 |
+
theme=gr.themes.Soft(primary_hue="violet", secondary_hue="violet", neutral_hue="gray",font=[gr.themes.GoogleFont("Exo"), "ui-sans-serif", "system-ui", "sans-serif"]).set(
|
100 |
+
body_background_fill_dark="#16141c",
|
101 |
+
block_background_fill_dark="#16141c",
|
102 |
+
block_title_background_fill_dark="#1e1c26",
|
103 |
+
input_background_fill_dark="#292733",
|
104 |
+
button_secondary_background_fill_dark="#24212b",
|
105 |
+
border_color_primary_dark="#343140",
|
106 |
+
background_fill_secondary_dark="#16141c"
|
107 |
+
),
|
108 |
+
css=css,
|
109 |
+
retry_btn="Retry",
|
110 |
+
undo_btn="Undo",
|
111 |
+
clear_btn="Clear",
|
112 |
+
submit_btn="Send",
|
113 |
+
description="Llama-cpp-agent: Chat multi llm selection"
|
114 |
)
|
115 |
|
|
|
116 |
if __name__ == "__main__":
|
117 |
+
demo.launch()
|