feat: setup local Qwen2 0.5
Browse files- app.py +9 -23
- requirements.txt +2 -1
app.py
CHANGED
@@ -1,10 +1,8 @@
|
|
1 |
import gradio as gr
|
2 |
-
from
|
3 |
|
4 |
-
""
|
5 |
-
|
6 |
-
"""
|
7 |
-
client = InferenceClient(model="meta-llama/Meta-Llama-3-8B-Instruct")
|
8 |
|
9 |
|
10 |
def respond(
|
@@ -25,19 +23,13 @@ def respond(
|
|
25 |
|
26 |
messages.append({"role": "user", "content": message})
|
27 |
|
28 |
-
response =
|
29 |
-
|
30 |
-
for message in client.chat_completion(
|
31 |
-
messages,
|
32 |
max_tokens=max_tokens,
|
33 |
-
stream=True,
|
34 |
temperature=temperature,
|
35 |
top_p=top_p,
|
36 |
-
)
|
37 |
-
|
38 |
-
|
39 |
-
response += token
|
40 |
-
yield response
|
41 |
|
42 |
|
43 |
"""
|
@@ -47,7 +39,7 @@ demo = gr.ChatInterface(
|
|
47 |
respond,
|
48 |
additional_inputs=[
|
49 |
gr.Textbox(
|
50 |
-
value="You are
|
51 |
label="System message",
|
52 |
),
|
53 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
@@ -60,13 +52,7 @@ demo = gr.ChatInterface(
|
|
60 |
label="Top-p (nucleus sampling)",
|
61 |
),
|
62 |
],
|
63 |
-
description=
|
64 |
-
examples=[
|
65 |
-
["پایتخت فرانسه کجاست؟"],
|
66 |
-
["دو بعلاوه دو چند میشود؟"],
|
67 |
-
["یک جک بگو"],
|
68 |
-
],
|
69 |
-
cache_examples=False,
|
70 |
)
|
71 |
|
72 |
|
|
|
1 |
import gradio as gr
|
2 |
+
from llama_cpp import Llama
|
3 |
|
4 |
+
model = "Qwen/Qwen1.5-0.5B-Chat-GGUF"
|
5 |
+
llm = Llama.from_pretrained(repo_id=model, filename="*q8_0.gguf", verbose=True)
|
|
|
|
|
6 |
|
7 |
|
8 |
def respond(
|
|
|
23 |
|
24 |
messages.append({"role": "user", "content": message})
|
25 |
|
26 |
+
response = llm.create_chat_completion(
|
27 |
+
messages=messages,
|
|
|
|
|
28 |
max_tokens=max_tokens,
|
|
|
29 |
temperature=temperature,
|
30 |
top_p=top_p,
|
31 |
+
)
|
32 |
+
return response
|
|
|
|
|
|
|
33 |
|
34 |
|
35 |
"""
|
|
|
39 |
respond,
|
40 |
additional_inputs=[
|
41 |
gr.Textbox(
|
42 |
+
value="You are a helpful assistant.",
|
43 |
label="System message",
|
44 |
),
|
45 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
|
|
52 |
label="Top-p (nucleus sampling)",
|
53 |
),
|
54 |
],
|
55 |
+
description=model,
|
|
|
|
|
|
|
|
|
|
|
|
|
56 |
)
|
57 |
|
58 |
|
requirements.txt
CHANGED
@@ -1 +1,2 @@
|
|
1 |
-
huggingface_hub==0.22.2
|
|
|
|
1 |
+
huggingface_hub==0.22.2
|
2 |
+
llama-cpp-python==0.2.78
|