Spaces:
Sleeping
Sleeping
gradio==5.1.0
Browse files- .python-version +1 -0
- README.md +1 -1
- app.py +23 -12
- pyproject.toml +20 -0
- requirements.txt +261 -8
- uv.lock +0 -0
.python-version
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
3.10
|
README.md
CHANGED
@@ -4,7 +4,7 @@ emoji: 🐢
|
|
4 |
colorFrom: purple
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
-
sdk_version:
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
---
|
|
|
4 |
colorFrom: purple
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
+
sdk_version: 5.1.0
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
---
|
app.py
CHANGED
@@ -7,7 +7,12 @@ from typing import Iterator
|
|
7 |
import gradio as gr
|
8 |
import spaces
|
9 |
import torch
|
10 |
-
from transformers import
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
DESCRIPTION = """# Swallow-13B instruct"""
|
13 |
|
@@ -18,7 +23,10 @@ if torch.cuda.is_available():
|
|
18 |
model_name = "tokyotech-llm/Swallow-13b-instruct-hf"
|
19 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
20 |
model = AutoModelForCausalLM.from_pretrained(
|
21 |
-
model_name,
|
|
|
|
|
|
|
22 |
)
|
23 |
|
24 |
MAX_INPUT_TOKENS = 2048
|
@@ -30,7 +38,9 @@ PROMPT_DICT = {
|
|
30 |
"### 指示:\n{instruction}\n\n### 入力:\n{input}\n\n### 応答:"
|
31 |
),
|
32 |
"prompt_no_input": (
|
33 |
-
"以下に、あるタスクを説明する指示があります。"
|
|
|
|
|
34 |
),
|
35 |
}
|
36 |
|
@@ -94,13 +104,8 @@ def process_example(instruction: str, input_text: str) -> Iterator[str]:
|
|
94 |
yield from run(instruction, input_text)
|
95 |
|
96 |
|
97 |
-
with gr.Blocks(
|
98 |
gr.Markdown(DESCRIPTION)
|
99 |
-
gr.DuplicateButton(
|
100 |
-
value="Duplicate Space for private use",
|
101 |
-
elem_id="duplicate-button",
|
102 |
-
visible=os.getenv("SHOW_DUPLICATE_BUTTON") == "1",
|
103 |
-
)
|
104 |
|
105 |
with gr.Row():
|
106 |
with gr.Column():
|
@@ -125,8 +130,14 @@ with gr.Blocks(css="style.css") as demo:
|
|
125 |
|
126 |
gr.Examples(
|
127 |
examples=[
|
128 |
-
[
|
129 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
130 |
["暴れん坊将軍って誰のことですか?", ""],
|
131 |
],
|
132 |
inputs=[instruction, input_text],
|
@@ -137,4 +148,4 @@ with gr.Blocks(css="style.css") as demo:
|
|
137 |
)
|
138 |
|
139 |
if __name__ == "__main__":
|
140 |
-
demo.
|
|
|
7 |
import gradio as gr
|
8 |
import spaces
|
9 |
import torch
|
10 |
+
from transformers import (
|
11 |
+
AutoModelForCausalLM,
|
12 |
+
AutoTokenizer,
|
13 |
+
BitsAndBytesConfig,
|
14 |
+
TextIteratorStreamer,
|
15 |
+
)
|
16 |
|
17 |
DESCRIPTION = """# Swallow-13B instruct"""
|
18 |
|
|
|
23 |
model_name = "tokyotech-llm/Swallow-13b-instruct-hf"
|
24 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
25 |
model = AutoModelForCausalLM.from_pretrained(
|
26 |
+
model_name,
|
27 |
+
quantization_config=BitsAndBytesConfig(load_in_8bit=True),
|
28 |
+
low_cpu_mem_usage=True,
|
29 |
+
device_map="auto",
|
30 |
)
|
31 |
|
32 |
MAX_INPUT_TOKENS = 2048
|
|
|
38 |
"### 指示:\n{instruction}\n\n### 入力:\n{input}\n\n### 応答:"
|
39 |
),
|
40 |
"prompt_no_input": (
|
41 |
+
"以下に、あるタスクを説明する指示があります。"
|
42 |
+
"リクエストを適切に完了するための回答を記述してください。\n\n"
|
43 |
+
"### 指示:\n{instruction}\n\n### 応答:"
|
44 |
),
|
45 |
}
|
46 |
|
|
|
104 |
yield from run(instruction, input_text)
|
105 |
|
106 |
|
107 |
+
with gr.Blocks(css_paths="style.css") as demo:
|
108 |
gr.Markdown(DESCRIPTION)
|
|
|
|
|
|
|
|
|
|
|
109 |
|
110 |
with gr.Row():
|
111 |
with gr.Column():
|
|
|
130 |
|
131 |
gr.Examples(
|
132 |
examples=[
|
133 |
+
[
|
134 |
+
"以下のトピックに関する詳細な情報を提供してください。",
|
135 |
+
"東京工業大学の主なキャンパスについて教えてください。",
|
136 |
+
],
|
137 |
+
[
|
138 |
+
"以下のトピックに関する詳細な情報を提供してください。",
|
139 |
+
"夢オチとは何かについて教えてください。",
|
140 |
+
],
|
141 |
["暴れん坊将軍って誰のことですか?", ""],
|
142 |
],
|
143 |
inputs=[instruction, input_text],
|
|
|
148 |
)
|
149 |
|
150 |
if __name__ == "__main__":
|
151 |
+
demo.launch()
|
pyproject.toml
ADDED
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
[project]
|
2 |
+
name = "swallow-13b-instruct"
|
3 |
+
version = "0.1.0"
|
4 |
+
description = ""
|
5 |
+
readme = "README.md"
|
6 |
+
requires-python = ">=3.10"
|
7 |
+
dependencies = [
|
8 |
+
"accelerate>=1.0.1",
|
9 |
+
"bitsandbytes>=0.44.1",
|
10 |
+
"blobfile>=3.0.0",
|
11 |
+
"gradio>=5.1.0",
|
12 |
+
"hf-transfer>=0.1.8",
|
13 |
+
"protobuf>=5.28.2",
|
14 |
+
"sentencepiece>=0.2.0",
|
15 |
+
"setuptools>=75.2.0",
|
16 |
+
"spaces>=0.30.4",
|
17 |
+
"tiktoken>=0.8.0",
|
18 |
+
"torch==2.4.0",
|
19 |
+
"transformers>=4.45.2",
|
20 |
+
]
|
requirements.txt
CHANGED
@@ -1,8 +1,261 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# This file was autogenerated by uv via the following command:
|
2 |
+
# uv pip compile pyproject.toml -o requirements.txt
|
3 |
+
accelerate==1.0.1
|
4 |
+
# via swallow-13b-instruct (pyproject.toml)
|
5 |
+
aiofiles==23.2.1
|
6 |
+
# via gradio
|
7 |
+
annotated-types==0.7.0
|
8 |
+
# via pydantic
|
9 |
+
anyio==4.6.2.post1
|
10 |
+
# via
|
11 |
+
# gradio
|
12 |
+
# httpx
|
13 |
+
# starlette
|
14 |
+
bitsandbytes==0.44.1
|
15 |
+
# via swallow-13b-instruct (pyproject.toml)
|
16 |
+
blobfile==3.0.0
|
17 |
+
# via swallow-13b-instruct (pyproject.toml)
|
18 |
+
certifi==2024.8.30
|
19 |
+
# via
|
20 |
+
# httpcore
|
21 |
+
# httpx
|
22 |
+
# requests
|
23 |
+
charset-normalizer==3.4.0
|
24 |
+
# via requests
|
25 |
+
click==8.1.7
|
26 |
+
# via
|
27 |
+
# typer
|
28 |
+
# uvicorn
|
29 |
+
exceptiongroup==1.2.2
|
30 |
+
# via anyio
|
31 |
+
fastapi==0.115.2
|
32 |
+
# via gradio
|
33 |
+
ffmpy==0.4.0
|
34 |
+
# via gradio
|
35 |
+
filelock==3.16.1
|
36 |
+
# via
|
37 |
+
# blobfile
|
38 |
+
# huggingface-hub
|
39 |
+
# torch
|
40 |
+
# transformers
|
41 |
+
# triton
|
42 |
+
fsspec==2024.9.0
|
43 |
+
# via
|
44 |
+
# gradio-client
|
45 |
+
# huggingface-hub
|
46 |
+
# torch
|
47 |
+
gradio==5.1.0
|
48 |
+
# via
|
49 |
+
# swallow-13b-instruct (pyproject.toml)
|
50 |
+
# spaces
|
51 |
+
gradio-client==1.4.0
|
52 |
+
# via gradio
|
53 |
+
h11==0.14.0
|
54 |
+
# via
|
55 |
+
# httpcore
|
56 |
+
# uvicorn
|
57 |
+
hf-transfer==0.1.8
|
58 |
+
# via swallow-13b-instruct (pyproject.toml)
|
59 |
+
httpcore==1.0.6
|
60 |
+
# via httpx
|
61 |
+
httpx==0.27.2
|
62 |
+
# via
|
63 |
+
# gradio
|
64 |
+
# gradio-client
|
65 |
+
# spaces
|
66 |
+
huggingface-hub==0.26.0
|
67 |
+
# via
|
68 |
+
# accelerate
|
69 |
+
# gradio
|
70 |
+
# gradio-client
|
71 |
+
# tokenizers
|
72 |
+
# transformers
|
73 |
+
idna==3.10
|
74 |
+
# via
|
75 |
+
# anyio
|
76 |
+
# httpx
|
77 |
+
# requests
|
78 |
+
jinja2==3.1.4
|
79 |
+
# via
|
80 |
+
# gradio
|
81 |
+
# torch
|
82 |
+
lxml==5.3.0
|
83 |
+
# via blobfile
|
84 |
+
markdown-it-py==3.0.0
|
85 |
+
# via rich
|
86 |
+
markupsafe==2.1.5
|
87 |
+
# via
|
88 |
+
# gradio
|
89 |
+
# jinja2
|
90 |
+
mdurl==0.1.2
|
91 |
+
# via markdown-it-py
|
92 |
+
mpmath==1.3.0
|
93 |
+
# via sympy
|
94 |
+
networkx==3.4.1
|
95 |
+
# via torch
|
96 |
+
numpy==2.1.2
|
97 |
+
# via
|
98 |
+
# accelerate
|
99 |
+
# bitsandbytes
|
100 |
+
# gradio
|
101 |
+
# pandas
|
102 |
+
# transformers
|
103 |
+
nvidia-cublas-cu12==12.1.3.1
|
104 |
+
# via
|
105 |
+
# nvidia-cudnn-cu12
|
106 |
+
# nvidia-cusolver-cu12
|
107 |
+
# torch
|
108 |
+
nvidia-cuda-cupti-cu12==12.1.105
|
109 |
+
# via torch
|
110 |
+
nvidia-cuda-nvrtc-cu12==12.1.105
|
111 |
+
# via torch
|
112 |
+
nvidia-cuda-runtime-cu12==12.1.105
|
113 |
+
# via torch
|
114 |
+
nvidia-cudnn-cu12==9.1.0.70
|
115 |
+
# via torch
|
116 |
+
nvidia-cufft-cu12==11.0.2.54
|
117 |
+
# via torch
|
118 |
+
nvidia-curand-cu12==10.3.2.106
|
119 |
+
# via torch
|
120 |
+
nvidia-cusolver-cu12==11.4.5.107
|
121 |
+
# via torch
|
122 |
+
nvidia-cusparse-cu12==12.1.0.106
|
123 |
+
# via
|
124 |
+
# nvidia-cusolver-cu12
|
125 |
+
# torch
|
126 |
+
nvidia-nccl-cu12==2.20.5
|
127 |
+
# via torch
|
128 |
+
nvidia-nvjitlink-cu12==12.6.77
|
129 |
+
# via
|
130 |
+
# nvidia-cusolver-cu12
|
131 |
+
# nvidia-cusparse-cu12
|
132 |
+
nvidia-nvtx-cu12==12.1.105
|
133 |
+
# via torch
|
134 |
+
orjson==3.10.9
|
135 |
+
# via gradio
|
136 |
+
packaging==24.1
|
137 |
+
# via
|
138 |
+
# accelerate
|
139 |
+
# gradio
|
140 |
+
# gradio-client
|
141 |
+
# huggingface-hub
|
142 |
+
# spaces
|
143 |
+
# transformers
|
144 |
+
pandas==2.2.3
|
145 |
+
# via gradio
|
146 |
+
pillow==10.4.0
|
147 |
+
# via gradio
|
148 |
+
protobuf==5.28.2
|
149 |
+
# via swallow-13b-instruct (pyproject.toml)
|
150 |
+
psutil==5.9.8
|
151 |
+
# via
|
152 |
+
# accelerate
|
153 |
+
# spaces
|
154 |
+
pycryptodomex==3.21.0
|
155 |
+
# via blobfile
|
156 |
+
pydantic==2.9.2
|
157 |
+
# via
|
158 |
+
# fastapi
|
159 |
+
# gradio
|
160 |
+
# spaces
|
161 |
+
pydantic-core==2.23.4
|
162 |
+
# via pydantic
|
163 |
+
pydub==0.25.1
|
164 |
+
# via gradio
|
165 |
+
pygments==2.18.0
|
166 |
+
# via rich
|
167 |
+
python-dateutil==2.9.0.post0
|
168 |
+
# via pandas
|
169 |
+
python-multipart==0.0.12
|
170 |
+
# via gradio
|
171 |
+
pytz==2024.2
|
172 |
+
# via pandas
|
173 |
+
pyyaml==6.0.2
|
174 |
+
# via
|
175 |
+
# accelerate
|
176 |
+
# gradio
|
177 |
+
# huggingface-hub
|
178 |
+
# transformers
|
179 |
+
regex==2024.9.11
|
180 |
+
# via
|
181 |
+
# tiktoken
|
182 |
+
# transformers
|
183 |
+
requests==2.32.3
|
184 |
+
# via
|
185 |
+
# huggingface-hub
|
186 |
+
# spaces
|
187 |
+
# tiktoken
|
188 |
+
# transformers
|
189 |
+
rich==13.9.2
|
190 |
+
# via typer
|
191 |
+
ruff==0.7.0
|
192 |
+
# via gradio
|
193 |
+
safetensors==0.4.5
|
194 |
+
# via
|
195 |
+
# accelerate
|
196 |
+
# transformers
|
197 |
+
semantic-version==2.10.0
|
198 |
+
# via gradio
|
199 |
+
sentencepiece==0.2.0
|
200 |
+
# via swallow-13b-instruct (pyproject.toml)
|
201 |
+
setuptools==75.2.0
|
202 |
+
# via swallow-13b-instruct (pyproject.toml)
|
203 |
+
shellingham==1.5.4
|
204 |
+
# via typer
|
205 |
+
six==1.16.0
|
206 |
+
# via python-dateutil
|
207 |
+
sniffio==1.3.1
|
208 |
+
# via
|
209 |
+
# anyio
|
210 |
+
# httpx
|
211 |
+
spaces==0.30.4
|
212 |
+
# via swallow-13b-instruct (pyproject.toml)
|
213 |
+
starlette==0.40.0
|
214 |
+
# via fastapi
|
215 |
+
sympy==1.13.3
|
216 |
+
# via torch
|
217 |
+
tiktoken==0.8.0
|
218 |
+
# via swallow-13b-instruct (pyproject.toml)
|
219 |
+
tokenizers==0.20.1
|
220 |
+
# via transformers
|
221 |
+
tomlkit==0.12.0
|
222 |
+
# via gradio
|
223 |
+
torch==2.4.0
|
224 |
+
# via
|
225 |
+
# swallow-13b-instruct (pyproject.toml)
|
226 |
+
# accelerate
|
227 |
+
# bitsandbytes
|
228 |
+
tqdm==4.66.5
|
229 |
+
# via
|
230 |
+
# huggingface-hub
|
231 |
+
# transformers
|
232 |
+
transformers==4.45.2
|
233 |
+
# via swallow-13b-instruct (pyproject.toml)
|
234 |
+
triton==3.0.0
|
235 |
+
# via torch
|
236 |
+
typer==0.12.5
|
237 |
+
# via gradio
|
238 |
+
typing-extensions==4.12.2
|
239 |
+
# via
|
240 |
+
# anyio
|
241 |
+
# fastapi
|
242 |
+
# gradio
|
243 |
+
# gradio-client
|
244 |
+
# huggingface-hub
|
245 |
+
# pydantic
|
246 |
+
# pydantic-core
|
247 |
+
# rich
|
248 |
+
# spaces
|
249 |
+
# torch
|
250 |
+
# typer
|
251 |
+
# uvicorn
|
252 |
+
tzdata==2024.2
|
253 |
+
# via pandas
|
254 |
+
urllib3==2.2.3
|
255 |
+
# via
|
256 |
+
# blobfile
|
257 |
+
# requests
|
258 |
+
uvicorn==0.32.0
|
259 |
+
# via gradio
|
260 |
+
websockets==12.0
|
261 |
+
# via gradio-client
|
uv.lock
ADDED
The diff for this file is too large to render.
See raw diff
|
|