Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -8,22 +8,26 @@ from llama_cpp_agent.chat_history import BasicChatHistory
|
|
8 |
from llama_cpp_agent.chat_history.messages import Roles
|
9 |
import gradio as gr
|
10 |
from huggingface_hub import hf_hub_download
|
|
|
11 |
repoId = "SakuraLLM/Sakura-14B-Qwen2beta-v0.9.2-GGUF"
|
12 |
filename = "sakura-14b-qwen2beta-v0.9.2-q6k.gguf"
|
13 |
-
|
|
|
14 |
hf_hub_download(
|
15 |
repo_id=repoId,
|
16 |
filename=filename,
|
17 |
local_dir="./models"
|
18 |
)
|
|
|
19 |
llm = None
|
20 |
llm_model = None
|
|
|
21 |
@spaces.GPU(duration=120)
|
22 |
def respond(
|
23 |
message,
|
24 |
history: list[tuple[str, str]],
|
25 |
model=filename,
|
26 |
-
system_message="
|
27 |
max_tokens=4096,
|
28 |
temperature=0.1,
|
29 |
top_p=0.3,
|
@@ -43,6 +47,7 @@ def respond(
|
|
43 |
n_ctx=8192,
|
44 |
)
|
45 |
llm_model = model
|
|
|
46 |
provider = LlamaCppPythonProvider(llm)
|
47 |
agent = LlamaCppAgent(
|
48 |
provider,
|
@@ -52,10 +57,9 @@ def respond(
|
|
52 |
)
|
53 |
|
54 |
settings = provider.get_provider_default_settings()
|
55 |
-
settings.temperature =
|
56 |
-
settings.top_k =
|
57 |
-
|
58 |
-
#settings.max_tokens = max_tokens
|
59 |
settings.repeat_penalty = repeat_penalty
|
60 |
settings.stream = True
|
61 |
messages = BasicChatHistory()
|
@@ -63,7 +67,7 @@ def respond(
|
|
63 |
for msn in history:
|
64 |
user = {
|
65 |
'role': Roles.user,
|
66 |
-
'content': "
|
67 |
}
|
68 |
assistant = {
|
69 |
'role': Roles.assistant,
|
@@ -85,6 +89,7 @@ def respond(
|
|
85 |
outputs += output
|
86 |
print(outputs)
|
87 |
|
|
|
88 |
# 檢查字串 '你是一个轻小说翻译模型' 是否在 outputs 中
|
89 |
if '你是一个轻小说翻译模型' in outputs:
|
90 |
a = outputs.index('你是一个轻小说翻译模型')
|
@@ -109,6 +114,7 @@ description = """<p align="center">Defaults to Sakura-14B-Qwen2beta</p>
|
|
109 |
<a href="https://huggingface.co/SakuraLLM/Sakura-14B-Qwen2beta-v0.9.2-GGUF" target="_blank">[Sakura-14B-Qwen2beta Model]</a>
|
110 |
</center></p>
|
111 |
"""
|
|
|
112 |
demo = gr.ChatInterface(
|
113 |
respond,
|
114 |
retry_btn="Retry",
|
@@ -123,6 +129,6 @@ demo = gr.ChatInterface(
|
|
123 |
show_copy_button=True
|
124 |
)
|
125 |
)
|
|
|
126 |
if __name__ == "__main__":
|
127 |
demo.launch()
|
128 |
-
|
|
|
8 |
from llama_cpp_agent.chat_history.messages import Roles
|
9 |
import gradio as gr
|
10 |
from huggingface_hub import hf_hub_download
|
11 |
+
|
12 |
repoId = "SakuraLLM/Sakura-14B-Qwen2beta-v0.9.2-GGUF"
|
13 |
filename = "sakura-14b-qwen2beta-v0.9.2-q6k.gguf"
|
14 |
+
|
15 |
+
# 下載 Sakura-14B 模型
|
16 |
hf_hub_download(
|
17 |
repo_id=repoId,
|
18 |
filename=filename,
|
19 |
local_dir="./models"
|
20 |
)
|
21 |
+
|
22 |
llm = None
|
23 |
llm_model = None
|
24 |
+
|
25 |
@spaces.GPU(duration=120)
|
26 |
def respond(
|
27 |
message,
|
28 |
history: list[tuple[str, str]],
|
29 |
model=filename,
|
30 |
+
system_message="你是一個專業的翻譯模型,能夠將日文翻譯成流暢的中文,並保持語境一致。",
|
31 |
max_tokens=4096,
|
32 |
temperature=0.1,
|
33 |
top_p=0.3,
|
|
|
47 |
n_ctx=8192,
|
48 |
)
|
49 |
llm_model = model
|
50 |
+
|
51 |
provider = LlamaCppPythonProvider(llm)
|
52 |
agent = LlamaCppAgent(
|
53 |
provider,
|
|
|
57 |
)
|
58 |
|
59 |
settings = provider.get_provider_default_settings()
|
60 |
+
settings.temperature = temperature
|
61 |
+
settings.top_k = top_k
|
62 |
+
settings.top_p = top_p
|
|
|
63 |
settings.repeat_penalty = repeat_penalty
|
64 |
settings.stream = True
|
65 |
messages = BasicChatHistory()
|
|
|
67 |
for msn in history:
|
68 |
user = {
|
69 |
'role': Roles.user,
|
70 |
+
'content': "請將以下日文文本翻譯成中文:\n" + msn[0]
|
71 |
}
|
72 |
assistant = {
|
73 |
'role': Roles.assistant,
|
|
|
89 |
outputs += output
|
90 |
print(outputs)
|
91 |
|
92 |
+
# 如果需要進行特定字串的移除或替換,可以在這裡調整
|
93 |
# 檢查字串 '你是一个轻小说翻译模型' 是否在 outputs 中
|
94 |
if '你是一个轻小说翻译模型' in outputs:
|
95 |
a = outputs.index('你是一个轻小说翻译模型')
|
|
|
114 |
<a href="https://huggingface.co/SakuraLLM/Sakura-14B-Qwen2beta-v0.9.2-GGUF" target="_blank">[Sakura-14B-Qwen2beta Model]</a>
|
115 |
</center></p>
|
116 |
"""
|
117 |
+
|
118 |
demo = gr.ChatInterface(
|
119 |
respond,
|
120 |
retry_btn="Retry",
|
|
|
129 |
show_copy_button=True
|
130 |
)
|
131 |
)
|
132 |
+
|
133 |
if __name__ == "__main__":
|
134 |
demo.launch()
|
|