sifujohn commited on
Commit
d2e98a1
1 Parent(s): dd4125a

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +103 -0
app.py ADDED
@@ -0,0 +1,103 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from huggingface_hub import InferenceClient
2
+ import gradio as gr
3
+
4
+ client = InferenceClient(
5
+ "models/google/gemma-7b-it"
6
+ )
7
+
8
+
9
+ def format_prompt(message, history):
10
+ prompt = "<s>"
11
+ for user_prompt, bot_response in history:
12
+ prompt += f"[INST] {user_prompt} [/INST]"
13
+ prompt += f" {bot_response}</s> "
14
+ prompt += f"[INST] {message} [/INST]"
15
+ return prompt
16
+
17
+ def generate(
18
+ prompt, history, system_prompt, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
19
+ ):
20
+ temperature = float(temperature)
21
+ if temperature < 1e-2:
22
+ temperature = 1e-2
23
+ top_p = float(top_p)
24
+
25
+ generate_kwargs = dict(
26
+ temperature=temperature,
27
+ max_new_tokens=max_new_tokens,
28
+ top_p=top_p,
29
+ repetition_penalty=repetition_penalty,
30
+ do_sample=True,
31
+ seed=42,
32
+ )
33
+
34
+ formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", history)
35
+ stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
36
+ output = ""
37
+
38
+ for response in stream:
39
+ output += response.token.text
40
+ yield output
41
+ return output
42
+
43
+
44
+ additional_inputs=[
45
+ gr.Textbox(
46
+ label="System Prompt",
47
+ max_lines=1,
48
+ interactive=True,
49
+ ),
50
+ gr.Slider(
51
+ label="Temperature",
52
+ value=0.9,
53
+ minimum=0.0,
54
+ maximum=1.0,
55
+ step=0.05,
56
+ interactive=True,
57
+ info="Higher values produce more diverse outputs",
58
+ ),
59
+ gr.Slider(
60
+ label="Max new tokens",
61
+ value=256,
62
+ minimum=0,
63
+ maximum=1048,
64
+ step=64,
65
+ interactive=True,
66
+ info="The maximum numbers of new tokens",
67
+ ),
68
+ gr.Slider(
69
+ label="Top-p (nucleus sampling)",
70
+ value=0.90,
71
+ minimum=0.0,
72
+ maximum=1,
73
+ step=0.05,
74
+ interactive=True,
75
+ info="Higher values sample more low-probability tokens",
76
+ ),
77
+ gr.Slider(
78
+ label="Repetition penalty",
79
+ value=1.2,
80
+ minimum=1.0,
81
+ maximum=2.0,
82
+ step=0.05,
83
+ interactive=True,
84
+ info="Penalize repeated tokens",
85
+ )
86
+ ]
87
+
88
+ examples=[["I'm planning a vacation to Japan. Can you suggest a one-week itinerary including must-visit places and local cuisines to try?", None, None, None, None, None, ],
89
+ ["Can you write a short story about a time-traveling detective who solves historical mysteries?", None, None, None, None, None,],
90
+ ["I'm trying to learn French. Can you provide some common phrases that would be useful for a beginner, along with their pronunciations?", None, None, None, None, None,],
91
+ ["I have chicken, rice, and bell peppers in my kitchen. Can you suggest an easy recipe I can make with these ingredients?", None, None, None, None, None,],
92
+ ["Can you explain how the QuickSort algorithm works and provide a Python implementation?", None, None, None, None, None,],
93
+ ["What are some unique features of Rust that make it stand out compared to other systems programming languages like C++?", None, None, None, None, None,],
94
+ ]
95
+
96
+ gr.ChatInterface(
97
+ fn=generate,
98
+ chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
99
+ additional_inputs=additional_inputs,
100
+ title="Gemma 7b",
101
+ examples=examples,
102
+ concurrency_limit=20,
103
+ ).launch(show_api=False)