Spaces:
Running
on
Zero
Running
on
Zero
tori29umai
commited on
Commit
•
82f2d09
1
Parent(s):
90c6cb4
Update app.py
Browse files
app.py
CHANGED
@@ -1,8 +1,10 @@
|
|
|
|
|
|
|
|
1 |
import spaces
|
2 |
import gradio as gr
|
3 |
from jinja2 import Template
|
4 |
from llama_cpp import Llama
|
5 |
-
import os
|
6 |
import configparser
|
7 |
from utils.dl_utils import dl_guff_model
|
8 |
|
@@ -61,9 +63,15 @@ class LlamaCppAdapter:
|
|
61 |
@spaces.GPU(duration=120)
|
62 |
def __init__(self, model_path, n_ctx=4096):
|
63 |
print(f"モデルの初期化: {model_path}")
|
64 |
-
|
|
|
|
|
|
|
|
|
65 |
|
66 |
def generate(self, prompt, max_new_tokens=4096, temperature=0.5, top_p=0.7, top_k=80, stop=["<END>"]):
|
|
|
|
|
67 |
return self._generate(prompt, max_new_tokens, temperature, top_p, top_k, stop)
|
68 |
|
69 |
def _generate(self, prompt: str, max_new_tokens: int, temperature: float, top_p: float, top_k: int, stop: list):
|
@@ -116,8 +124,12 @@ class CharacterMaker:
|
|
116 |
prompt = self._generate_aki(input_str)
|
117 |
print(prompt)
|
118 |
print("-----------------")
|
119 |
-
|
120 |
-
|
|
|
|
|
|
|
|
|
121 |
self.history.append({"user": input_str, "assistant": res_text})
|
122 |
return res_text
|
123 |
|
@@ -139,7 +151,6 @@ class CharacterMaker:
|
|
139 |
self.settings["example_quotes"],
|
140 |
input_str
|
141 |
)
|
142 |
-
print(prompt)
|
143 |
return prompt
|
144 |
|
145 |
def update_settings(self, new_settings):
|
@@ -193,7 +204,6 @@ function adjustChatbotHeight() {
|
|
193 |
chatbot.style.height = window.innerHeight * 0.6 + 'px';
|
194 |
}
|
195 |
}
|
196 |
-
|
197 |
// ページ読み込み時と画面サイズ変更時にチャットボットの高さを調整
|
198 |
window.addEventListener('load', adjustChatbotHeight);
|
199 |
window.addEventListener('resize', adjustChatbotHeight);
|
|
|
1 |
+
import os
|
2 |
+
os.environ['CUDA_VISIBLE_DEVICES'] = ''
|
3 |
+
|
4 |
import spaces
|
5 |
import gradio as gr
|
6 |
from jinja2 import Template
|
7 |
from llama_cpp import Llama
|
|
|
8 |
import configparser
|
9 |
from utils.dl_utils import dl_guff_model
|
10 |
|
|
|
63 |
@spaces.GPU(duration=120)
|
64 |
def __init__(self, model_path, n_ctx=4096):
|
65 |
print(f"モデルの初期化: {model_path}")
|
66 |
+
try:
|
67 |
+
self.llama = Llama(model_path=model_path, n_ctx=n_ctx, n_gpu_layers=0)
|
68 |
+
except Exception as e:
|
69 |
+
print(f"モデルの初期化中にエラーが発生しました: {e}")
|
70 |
+
self.llama = None
|
71 |
|
72 |
def generate(self, prompt, max_new_tokens=4096, temperature=0.5, top_p=0.7, top_k=80, stop=["<END>"]):
|
73 |
+
if self.llama is None:
|
74 |
+
return {"choices": [{"text": "モデルの初期化に失敗しました。"}]}
|
75 |
return self._generate(prompt, max_new_tokens, temperature, top_p, top_k, stop)
|
76 |
|
77 |
def _generate(self, prompt: str, max_new_tokens: int, temperature: float, top_p: float, top_k: int, stop: list):
|
|
|
124 |
prompt = self._generate_aki(input_str)
|
125 |
print(prompt)
|
126 |
print("-----------------")
|
127 |
+
try:
|
128 |
+
res = self.llama.generate(prompt, max_new_tokens=1000, stop=["<END>", "\n"])
|
129 |
+
res_text = res["choices"][0]["text"]
|
130 |
+
except Exception as e:
|
131 |
+
print(f"生成中にエラーが発生しました: {e}")
|
132 |
+
res_text = "申し訳ありません。応答の生成中にエラーが発生しました。"
|
133 |
self.history.append({"user": input_str, "assistant": res_text})
|
134 |
return res_text
|
135 |
|
|
|
151 |
self.settings["example_quotes"],
|
152 |
input_str
|
153 |
)
|
|
|
154 |
return prompt
|
155 |
|
156 |
def update_settings(self, new_settings):
|
|
|
204 |
chatbot.style.height = window.innerHeight * 0.6 + 'px';
|
205 |
}
|
206 |
}
|
|
|
207 |
// ページ読み込み時と画面サイズ変更時にチャットボットの高さを調整
|
208 |
window.addEventListener('load', adjustChatbotHeight);
|
209 |
window.addEventListener('resize', adjustChatbotHeight);
|