Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -5,6 +5,16 @@ import torch
|
|
5 |
import gradio as gr
|
6 |
from threading import Thread
|
7 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
9 |
|
10 |
TITLE = "<h1><center>Mistral-lab</center></h1>"
|
@@ -15,25 +25,32 @@ PLACEHOLDER = """
|
|
15 |
</center>
|
16 |
"""
|
17 |
|
18 |
-
|
19 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
|
|
|
|
|
21 |
mistral_models_path = Path.home().joinpath('mistral_models', '8B-Instruct')
|
22 |
mistral_models_path.mkdir(parents=True, exist_ok=True)
|
23 |
|
24 |
snapshot_download(repo_id="mistralai/Ministral-8B-Instruct-2410", allow_patterns=["params.json", "consolidated.safetensors", "tekken.json"], local_dir=mistral_models_path)
|
25 |
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
from mistral_common.tokens.tokenizers.mistral import MistralTokenizer
|
30 |
-
from mistral_common.protocol.instruct.messages import AssistantMessage, UserMessage
|
31 |
-
from mistral_common.protocol.instruct.request import ChatCompletionRequest
|
32 |
-
|
33 |
-
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
34 |
-
|
35 |
tokenizer = MistralTokenizer.from_file(f"{mistral_models_path}/tekken.json")
|
36 |
-
model = Transformer.from_folder(
|
|
|
|
|
|
|
37 |
|
38 |
|
39 |
@spaces.GPU()
|
@@ -64,12 +81,23 @@ def stream_chat(
|
|
64 |
eos_id=tokenizer.instruct_tokenizer.tokenizer.eos_id)
|
65 |
|
66 |
result = tokenizer.instruct_tokenizer.tokenizer.decode(out_tokens[0])
|
67 |
-
|
68 |
-
|
|
|
|
|
69 |
|
70 |
-
chatbot = gr.Chatbot(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
|
72 |
-
with gr.Blocks(theme="ocean") as demo:
|
73 |
gr.HTML(TITLE)
|
74 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
75 |
gr.ChatInterface(
|
@@ -95,13 +123,6 @@ with gr.Blocks(theme="ocean") as demo:
|
|
95 |
render=False,
|
96 |
),
|
97 |
],
|
98 |
-
examples=[
|
99 |
-
["Help me study vocabulary: write a sentence for me to fill in the blank, and I'll try to pick the correct option."],
|
100 |
-
["What are 5 creative things I could do with my kids' art? I don't want to throw them away, but it's also so much clutter."],
|
101 |
-
["Tell me a random fun fact about the Roman Empire."],
|
102 |
-
["Show me a code snippet of a website's sticky header in CSS and JavaScript."],
|
103 |
-
],
|
104 |
-
cache_examples=False,
|
105 |
)
|
106 |
|
107 |
|
|
|
5 |
import gradio as gr
|
6 |
from threading import Thread
|
7 |
|
8 |
+
from huggingface_hub import snapshot_download
|
9 |
+
from pathlib import Path
|
10 |
+
|
11 |
+
from mistral_inference.transformer import Transformer
|
12 |
+
from mistral_inference.generate import generate
|
13 |
+
|
14 |
+
from mistral_common.tokens.tokenizers.mistral import MistralTokenizer
|
15 |
+
from mistral_common.protocol.instruct.messages import AssistantMessage, UserMessage
|
16 |
+
from mistral_common.protocol.instruct.request import ChatCompletionRequest
|
17 |
+
|
18 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
19 |
|
20 |
TITLE = "<h1><center>Mistral-lab</center></h1>"
|
|
|
25 |
</center>
|
26 |
"""
|
27 |
|
28 |
+
CSS = """
|
29 |
+
.duplicate-button {
|
30 |
+
margin: auto !important;
|
31 |
+
color: white !important;
|
32 |
+
background: black !important;
|
33 |
+
border-radius: 100vh !important;
|
34 |
+
}
|
35 |
+
h3 {
|
36 |
+
text-align: center;
|
37 |
+
}
|
38 |
+
"""
|
39 |
|
40 |
+
|
41 |
+
# download model
|
42 |
mistral_models_path = Path.home().joinpath('mistral_models', '8B-Instruct')
|
43 |
mistral_models_path.mkdir(parents=True, exist_ok=True)
|
44 |
|
45 |
snapshot_download(repo_id="mistralai/Ministral-8B-Instruct-2410", allow_patterns=["params.json", "consolidated.safetensors", "tekken.json"], local_dir=mistral_models_path)
|
46 |
|
47 |
+
# tokenizer
|
48 |
+
device = "cuda" if torch.cuda.is_available() else "cpu" # for GPU usage or "cpu" for CPU usage
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
49 |
tokenizer = MistralTokenizer.from_file(f"{mistral_models_path}/tekken.json")
|
50 |
+
model = Transformer.from_folder(
|
51 |
+
mistral_models_path,
|
52 |
+
device=device,
|
53 |
+
dtype=torch.bfloat16)
|
54 |
|
55 |
|
56 |
@spaces.GPU()
|
|
|
81 |
eos_id=tokenizer.instruct_tokenizer.tokenizer.eos_id)
|
82 |
|
83 |
result = tokenizer.instruct_tokenizer.tokenizer.decode(out_tokens[0])
|
84 |
+
|
85 |
+
for i in range(len(result)):
|
86 |
+
time.sleep(0.05)
|
87 |
+
yield result[: i + 1]
|
88 |
|
89 |
+
chatbot = gr.Chatbot(
|
90 |
+
height=600,
|
91 |
+
placeholder=PLACEHOLDER,
|
92 |
+
examples=[
|
93 |
+
{"text": "Help me study vocabulary: write a sentence for me to fill in the blank, and I'll try to pick the correct option."},
|
94 |
+
{"text": "What are 5 creative things I could do with my kids' art? I don't want to throw them away, but it's also so much clutter."},
|
95 |
+
{"text": "Tell me a random fun fact about the Roman Empire."},
|
96 |
+
{"text": "Show me a code snippet of a website's sticky header in CSS and JavaScript."},
|
97 |
+
],
|
98 |
+
)
|
99 |
|
100 |
+
with gr.Blocks(theme="ocean", css=CSS) as demo:
|
101 |
gr.HTML(TITLE)
|
102 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
103 |
gr.ChatInterface(
|
|
|
123 |
render=False,
|
124 |
),
|
125 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
126 |
)
|
127 |
|
128 |
|