File size: 4,104 Bytes
f64354b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
import gradio as gr
from huggingface_hub import InferenceClient
import pytesseract
from pypdf import PdfReader
import ocrmypdf
from PIL import Image
import os

# Image to Text

def fn_image_to_text(input_image):
    return pytesseract.image_to_string(Image.open(input_image))

# PDF to Text

def fn_pdf_to_text(input_pdf):
    reader = PdfReader(input_pdf)
    
    output_pdf = ""
    for page in reader.pages:
        output_pdf+=page.extract_text()
  
    image_count = 0
    for page in reader.pages:
        image_count += len(page.images)

    if image_count > 0 and len(output_pdf) < 1000:
        input_pdf_ocr = input_pdf.replace(".pdf", " - OCR.pdf")
        ocrmypdf.ocr(input_pdf, input_pdf_ocr, force_ocr=True)
    
        reader = PdfReader(input_pdf_ocr)
        output_pdf = ""
        for page in reader.pages:
            output_pdf+=page.extract_text()

        os.remove(input_pdf_ocr)
  
    return output_pdf

# Inference

model_text = "google/gemma-3-27b-it"
#model_text = "google/gemma-2-27b-it"
#model_vision = "google/paligemma2-3b-pt-224"

client = InferenceClient()

def fn_text(
    prompt,
    history,
    input,
    #system_prompt,
    max_tokens,
    temperature,
    top_p,
):
    if input:
        if os.path.splitext(input)[1].lower() in [".png", ".jpg", ".jpeg"]:
            output = fn_image_to_text(input)
        if os.path.splitext(input)[1].lower() == ".pdf":
            output = fn_pdf_to_text(input)
    else:
        output = ""
    
    #messages = [{"role": "system", "content": system_prompt}]
    #history.append(messages[0])
    #messages.append({"role": "user", "content": prompt})
    #history.append(messages[1])

    messages = [{"role": "user", "content": prompt + " " + output}]
    history.append(messages[0])
    
    stream = client.chat.completions.create(
        model = model_text,
        messages = history,
        max_tokens = max_tokens,
        temperature = temperature,
        top_p = top_p,
        stream = True,
    )
    
    chunks = []
    for chunk in stream:
        chunks.append(chunk.choices[0].delta.content or "")
        yield "".join(chunks)

app_text = gr.ChatInterface(
    fn = fn_text,
    type = "messages",
    additional_inputs = [
        gr.File(type="filepath", label="Input"),
        #gr.Textbox(value="You are a helpful assistant.", label="System Prompt"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max Tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-P"),
    ],
    title = "Google Gemma",
    description = model_text,
)
"""
def fn_vision(
    prompt,
    image_url,
    #system_prompt,
    max_tokens,
    temperature,
    top_p,
):
    messages = [{"role": "user", "content": [{"type": "text", "text": prompt}]}]
    
    if image_url:
        messages[0]["content"].append({"type": "image_url", "image_url": {"url": image_url}})
    
    stream = client.chat.completions.create(
        model = model_vision,
        messages = messages,
        max_tokens = max_tokens,
        temperature = temperature,
        top_p = top_p,
        stream = True,
    )
    
    chunks = []
    for chunk in stream:
        chunks.append(chunk.choices[0].delta.content or "")
        yield "".join(chunks)

app_vision = gr.Interface(
    fn = fn_vision,
    inputs = [
        gr.Textbox(label="Prompt"),
        gr.Textbox(label="Image URL")
    ],
    outputs = [
        gr.Textbox(label="Output")
    ],
    additional_inputs = [
        #gr.Textbox(value="You are a helpful assistant.", label="System Prompt"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max Tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-P"),
    ],
    title = "Google Gemma",
    description = model_vision,
)
"""
app = gr.TabbedInterface(
    [app_text],
    ["Text"]
).launch()

#if __name__ == "__main__":
#    app.launch()