Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -2,23 +2,25 @@ import os
|
|
2 |
import time
|
3 |
import spaces
|
4 |
import torch
|
5 |
-
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
|
6 |
import gradio as gr
|
7 |
from threading import Thread
|
8 |
|
|
|
9 |
MODEL_LIST = ["mistralai/mathstral-7B-v0.1"]
|
10 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
11 |
MODEL = os.environ.get("MODEL_ID")
|
12 |
|
13 |
-
TITLE = "<h1><center>MathΣtral</center></h1>"
|
14 |
-
|
15 |
PLACEHOLDER = """
|
16 |
<center>
|
17 |
-
<p>MathΣtral -
|
|
|
|
|
|
|
|
|
18 |
</center>
|
19 |
"""
|
20 |
|
21 |
-
|
22 |
CSS = """
|
23 |
.duplicate-button {
|
24 |
margin: auto !important;
|
@@ -26,69 +28,87 @@ CSS = """
|
|
26 |
background: black !important;
|
27 |
border-radius: 100vh !important;
|
28 |
}
|
29 |
-
|
30 |
text-align: center;
|
|
|
|
|
31 |
}
|
32 |
"""
|
33 |
|
34 |
-
|
|
|
|
|
35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
tokenizer = AutoTokenizer.from_pretrained(MODEL)
|
37 |
model = AutoModelForCausalLM.from_pretrained(
|
38 |
MODEL,
|
39 |
torch_dtype=torch.bfloat16,
|
40 |
device_map="auto",
|
41 |
-
|
|
|
42 |
|
|
|
43 |
@spaces.GPU()
|
44 |
def stream_chat(
|
45 |
-
message: str,
|
46 |
-
history: list,
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
|
|
51 |
penalty: float = 1.2,
|
52 |
):
|
53 |
print(f'message: {message}')
|
54 |
print(f'history: {history}')
|
55 |
|
56 |
-
conversation
|
|
|
57 |
for prompt, answer in history:
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
])
|
62 |
|
63 |
-
|
|
|
64 |
|
65 |
-
input_text=tokenizer.apply_chat_template(conversation, tokenize=False)
|
66 |
-
inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
|
67 |
streamer = TextIteratorStreamer(tokenizer, timeout=60.0, skip_prompt=True, skip_special_tokens=True)
|
68 |
-
|
69 |
generate_kwargs = dict(
|
70 |
-
input_ids=
|
71 |
-
max_new_tokens
|
72 |
-
do_sample
|
73 |
-
top_p
|
74 |
-
top_k
|
75 |
-
temperature
|
|
|
76 |
streamer=streamer,
|
77 |
-
pad_token_id = 10,
|
78 |
)
|
79 |
|
80 |
with torch.no_grad():
|
81 |
thread = Thread(target=model.generate, kwargs=generate_kwargs)
|
82 |
thread.start()
|
83 |
-
|
84 |
buffer = ""
|
85 |
for new_text in streamer:
|
86 |
buffer += new_text
|
87 |
-
|
|
|
|
|
88 |
|
89 |
-
|
90 |
-
chatbot = gr.Chatbot(height=
|
91 |
|
|
|
92 |
footer = """
|
93 |
<div style="text-align: center; margin-top: 20px;">
|
94 |
<a href="https://www.linkedin.com/in/pejman-ebrahimi-4a60151a7/" target="_blank">LinkedIn</a> |
|
@@ -99,6 +119,7 @@ footer = """
|
|
99 |
</div>
|
100 |
"""
|
101 |
|
|
|
102 |
with gr.Blocks(css=CSS, theme="Ajaxon6255/Emerald_Isle") as demo:
|
103 |
gr.HTML(TITLE)
|
104 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
@@ -108,11 +129,16 @@ with gr.Blocks(css=CSS, theme="Ajaxon6255/Emerald_Isle") as demo:
|
|
108 |
fill_height=True,
|
109 |
additional_inputs_accordion=gr.Accordion(label="��️ Parameters", open=False, render=False),
|
110 |
additional_inputs=[
|
|
|
|
|
|
|
|
|
|
|
111 |
gr.Slider(
|
112 |
minimum=0,
|
113 |
maximum=1,
|
114 |
step=0.1,
|
115 |
-
value=0.
|
116 |
label="Temperature",
|
117 |
render=False,
|
118 |
),
|
@@ -157,7 +183,8 @@ with gr.Blocks(css=CSS, theme="Ajaxon6255/Emerald_Isle") as demo:
|
|
157 |
],
|
158 |
cache_examples=False,
|
159 |
)
|
|
|
160 |
|
161 |
-
|
162 |
if __name__ == "__main__":
|
163 |
-
demo.launch()
|
|
|
2 |
import time
|
3 |
import spaces
|
4 |
import torch
|
5 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer, BitsAndBytesConfig
|
6 |
import gradio as gr
|
7 |
from threading import Thread
|
8 |
|
9 |
+
# Define constants and configuration
|
10 |
MODEL_LIST = ["mistralai/mathstral-7B-v0.1"]
|
11 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
12 |
MODEL = os.environ.get("MODEL_ID")
|
13 |
|
|
|
|
|
14 |
PLACEHOLDER = """
|
15 |
<center>
|
16 |
+
<p>MathΣtral - Your Math advisor</p>
|
17 |
+
<p>Hi! I'm MisMath. A Math advisor. My model is based on mathstral-7B-v0.1. Feel free to ask your questions</p>
|
18 |
+
<p>Mathstral 7B is a model specializing in mathematical and scientific tasks, based on Mistral 7B.</p>
|
19 |
+
<p>mathstral-7B-v0.1 is the first Mathstral model</p>
|
20 |
+
<img src="Mistral.png" alt="MathStral Model" style="width:300px;height:200px;">
|
21 |
</center>
|
22 |
"""
|
23 |
|
|
|
24 |
CSS = """
|
25 |
.duplicate-button {
|
26 |
margin: auto !important;
|
|
|
28 |
background: black !important;
|
29 |
border-radius: 100vh !important;
|
30 |
}
|
31 |
+
h1 {
|
32 |
text-align: center;
|
33 |
+
font-size: 2em;
|
34 |
+
color: #333;
|
35 |
}
|
36 |
"""
|
37 |
|
38 |
+
TITLE = "<h1><center>MathΣtral - Your Math advisor</center></h1>"
|
39 |
+
|
40 |
+
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
41 |
|
42 |
+
# Configuration for model quantization
|
43 |
+
quantization_config = BitsAndBytesConfig(
|
44 |
+
load_in_4bit=True,
|
45 |
+
bnb_4bit_compute_dtype=torch.bfloat16,
|
46 |
+
bnb_4bit_use_double_quant=True,
|
47 |
+
bnb_4bit_quant_type="nf4"
|
48 |
+
)
|
49 |
+
|
50 |
+
# Initialize tokenizer and model
|
51 |
tokenizer = AutoTokenizer.from_pretrained(MODEL)
|
52 |
model = AutoModelForCausalLM.from_pretrained(
|
53 |
MODEL,
|
54 |
torch_dtype=torch.bfloat16,
|
55 |
device_map="auto",
|
56 |
+
quantization_config=quantization_config
|
57 |
+
)
|
58 |
|
59 |
+
# Define the chat streaming function
|
60 |
@spaces.GPU()
|
61 |
def stream_chat(
|
62 |
+
message: str,
|
63 |
+
history: list,
|
64 |
+
system_prompt: str,
|
65 |
+
temperature: float = 0.8,
|
66 |
+
max_new_tokens: int = 1024,
|
67 |
+
top_p: float = 1.0,
|
68 |
+
top_k: int = 20,
|
69 |
penalty: float = 1.2,
|
70 |
):
|
71 |
print(f'message: {message}')
|
72 |
print(f'history: {history}')
|
73 |
|
74 |
+
# Prepare the conversation context
|
75 |
+
conversation_text = system_prompt + "\n"
|
76 |
for prompt, answer in history:
|
77 |
+
conversation_text += f"User: {prompt}\nAssistant: {answer}\n"
|
78 |
+
|
79 |
+
conversation_text += f"User: {message}\nAssistant:"
|
|
|
80 |
|
81 |
+
# Tokenize the conversation text
|
82 |
+
input_ids = tokenizer(conversation_text, return_tensors="pt").input_ids.to(model.device)
|
83 |
|
|
|
|
|
84 |
streamer = TextIteratorStreamer(tokenizer, timeout=60.0, skip_prompt=True, skip_special_tokens=True)
|
85 |
+
|
86 |
generate_kwargs = dict(
|
87 |
+
input_ids=input_ids,
|
88 |
+
max_new_tokens=max_new_tokens,
|
89 |
+
do_sample=False if temperature == 0 else True,
|
90 |
+
top_p=top_p,
|
91 |
+
top_k=top_k,
|
92 |
+
temperature=temperature,
|
93 |
+
eos_token_id=[128001, 128008, 128009],
|
94 |
streamer=streamer,
|
|
|
95 |
)
|
96 |
|
97 |
with torch.no_grad():
|
98 |
thread = Thread(target=model.generate, kwargs=generate_kwargs)
|
99 |
thread.start()
|
100 |
+
|
101 |
buffer = ""
|
102 |
for new_text in streamer:
|
103 |
buffer += new_text
|
104 |
+
# Clean the buffer to remove unwanted prefixes
|
105 |
+
cleaned_text = buffer.split("Assistant:")[-1].strip()
|
106 |
+
yield cleaned_text
|
107 |
|
108 |
+
# Define the Gradio chatbot component
|
109 |
+
chatbot = gr.Chatbot(height=500, placeholder=PLACEHOLDER)
|
110 |
|
111 |
+
# Define the footer with links
|
112 |
footer = """
|
113 |
<div style="text-align: center; margin-top: 20px;">
|
114 |
<a href="https://www.linkedin.com/in/pejman-ebrahimi-4a60151a7/" target="_blank">LinkedIn</a> |
|
|
|
119 |
</div>
|
120 |
"""
|
121 |
|
122 |
+
# Create and launch the Gradio interface
|
123 |
with gr.Blocks(css=CSS, theme="Ajaxon6255/Emerald_Isle") as demo:
|
124 |
gr.HTML(TITLE)
|
125 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
|
|
129 |
fill_height=True,
|
130 |
additional_inputs_accordion=gr.Accordion(label="��️ Parameters", open=False, render=False),
|
131 |
additional_inputs=[
|
132 |
+
gr.Textbox(
|
133 |
+
value="You are a helpful assistant for Math questions and complex calculations and programming and your name is MisMath",
|
134 |
+
label="System Prompt",
|
135 |
+
render=False,
|
136 |
+
),
|
137 |
gr.Slider(
|
138 |
minimum=0,
|
139 |
maximum=1,
|
140 |
step=0.1,
|
141 |
+
value=0.8,
|
142 |
label="Temperature",
|
143 |
render=False,
|
144 |
),
|
|
|
183 |
],
|
184 |
cache_examples=False,
|
185 |
)
|
186 |
+
gr.HTML(footer)
|
187 |
|
188 |
+
# Launch the application
|
189 |
if __name__ == "__main__":
|
190 |
+
demo.launch()
|