Duplicate from HamidRezaAttar/gpt2-home

Co-authored-by: HamidReza Fatollah Zadeh Attar <[email protected]>

.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,13 @@
+---
+title: Gpt2 Home
+emoji: 🏢
+colorFrom: pink
+colorTo: blue
+sdk: streamlit
+app_file: app.py
+pinned: false
+license: apache-2.0
+duplicated_from: HamidRezaAttar/gpt2-home
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces#reference

app.py

@@ -0,0 +1,140 @@
+import streamlit as st
+from transformers import pipeline, set_seed
+from transformers import AutoTokenizer
+from normalizer import Normalizer
+import random
+
+import meta
+import examples
+from utils import (
+    remote_css,
+    local_css
+)
+
+
+class TextGeneration:
+    def __init__(self):
+        self.debug = False
+        self.dummy_output = None
+        self.tokenizer = None
+        self.generator = None
+        self.task = "text-generation"
+        self.model_name_or_path = "HamidRezaAttar/gpt2-product-description-generator"
+        set_seed(42)
+
+    def load(self):
+        if not self.debug:
+            self.tokenizer = AutoTokenizer.from_pretrained(self.model_name_or_path)
+            self.generator = pipeline(self.task, model=self.model_name_or_path, tokenizer=self.model_name_or_path)
+
+    def generate(self, prompt, generation_kwargs):
+        if not self.debug:
+            generation_kwargs["num_return_sequences"] = 1
+
+            max_length = len(self.tokenizer(prompt)["input_ids"]) + generation_kwargs["max_length"]
+            generation_kwargs["max_length"] = max_length
+
+            generation_kwargs["return_full_text"] = False
+
+            return self.generator(
+                prompt,
+                **generation_kwargs,
+            )[0]["generated_text"]
+
+        return self.dummy_output
+
+
+@st.cache(allow_output_mutation=True)
+def load_text_generator():
+    generator = TextGeneration()
+    generator.load()
+    return generator
+
+
+def main():
+    st.set_page_config(
+        page_title="GPT2 - Home",
+        page_icon="🏡",
+        layout="wide",
+        initial_sidebar_state="expanded"
+    )
+    remote_css("https://fonts.googleapis.com/css2?family=Roboto:wght@300&display=swap%22%20rel=%22stylesheet%22")
+    local_css("assets/ltr.css")
+    generator = load_text_generator()
+
+    st.sidebar.markdown(meta.SIDEBAR_INFO)
+
+    max_length = st.sidebar.slider(
+        label='Max Length',
+        help="The maximum length of the sequence to be generated.",
+        min_value=1,
+        max_value=128,
+        value=50,
+        step=1
+    )
+    top_k = st.sidebar.slider(
+        label='Top-k',
+        help="The number of highest probability vocabulary tokens to keep for top-k-filtering",
+        min_value=40,
+        max_value=80,
+        value=50,
+        step=1
+    )
+    top_p = st.sidebar.slider(
+        label='Top-p',
+        help="Only the most probable tokens with probabilities that add up to `top_p` or higher are kept for "
+             "generation.",
+        min_value=0.0,
+        max_value=1.0,
+        value=0.95,
+        step=0.01
+    )
+    temperature = st.sidebar.slider(
+        label='Temperature',
+        help="The value used to module the next token probabilities",
+        min_value=0.1,
+        max_value=10.0,
+        value=1.0,
+        step=0.05
+    )
+    do_sample = st.sidebar.selectbox(
+        label='Sampling ?',
+        options=(True, False),
+        help="Whether or not to use sampling; use greedy decoding otherwise.",
+    )
+    generation_kwargs = {
+        "max_length": max_length,
+        "top_k": top_k,
+        "top_p": top_p,
+        "temperature": temperature,
+        "do_sample": do_sample,
+    }
+
+    st.markdown(meta.HEADER_INFO)
+    prompts = list(examples.EXAMPLES.keys()) + ["Custom"]
+    prompt = st.selectbox('Examples', prompts, index=len(prompts) - 1)
+
+    if prompt == "Custom":
+        prompt_box = meta.PROMPT_BOX
+    else:
+        prompt_box = random.choice(examples.EXAMPLES[prompt])
+
+    text = st.text_area("Enter text", prompt_box)
+    generation_kwargs_ph = st.empty()
+    cleaner = Normalizer()
+    if st.button("Generate !"):
+        with st.spinner(text="Generating ..."):
+            generation_kwargs_ph.markdown(", ".join([f"`{k}`: {v}" for k, v in generation_kwargs.items()]))
+            if text:
+                generated_text = generator.generate(text, generation_kwargs)
+                generated_text = cleaner.clean_txt(generated_text)
+                st.markdown(
+                    f'<p class="ltr ltr-box">'
+                    f'<span class="result-text">{text} <span>'
+                    f'<span class="result-text generated-text">{generated_text}</span>'
+                    f'</p>',
+                    unsafe_allow_html=True
+                )
+
+if __name__ == '__main__':
+    main()

assets/ltr.css

@@ -0,0 +1,22 @@
+.ltr,
+textarea {
+    font-family: Roboto !important;
+    text-align: left;
+    direction: ltr !important;
+}
+.ltr-box {
+    border-bottom: 1px solid #ddd;
+    padding-bottom: 20px;
+}
+.rtl {
+    text-align: left;
+    direction: ltr !important;
+}
+
+span.result-text {
+    padding: 3px 3px;
+    line-height: 32px;
+}
+span.generated-text {
+    background-color: rgb(118 200 147 / 13%);
+}

examples.py

@@ -0,0 +1,11 @@
+EXAMPLES = {
+    "Table": [
+        "Handcrafted of solid acacia in weathered gray, our round Jozy drop-leaf dining table is a space-saving."
+    ],
+    "Bed": [
+        "Maximize your bedroom space without sacrificing style with the storage bed."
+    ],
+    "Sofa": [
+        "Our plush and luxurious Emmett modular sofa brings custom comfort to your living space."
+    ]
+}

meta.py

@@ -0,0 +1,8 @@
+HEADER_INFO = """
+# GPT2 - Home
+English GPT-2 home product description generator demo.
+""".strip()
+SIDEBAR_INFO = """
+# Configuration
+""".strip()
+PROMPT_BOX = "Enter your text..."

normalizer.py

@@ -0,0 +1,17 @@
+class Normalizer:
+
+    def __init__(self):
+        pass
+
+    def remove_repetitions(self, text):
+        first_ocurrences = []
+        for sentence in text.split("."):
+            if sentence not in first_ocurrences:
+                first_ocurrences.append(sentence)
+        return '.'.join(first_ocurrences)
+
+    def trim_last_sentence(self, text):
+        return text[:text.rfind(".")+1]
+            
+    def clean_txt(self, text):
+        return self.trim_last_sentence(self.remove_repetitions(text))

requirements.txt

@@ -0,0 +1,4 @@
+torch
+Pillow==9.0.1
+streamlit==1.5.1
+transformers==4.16.2

utils.py

@@ -0,0 +1,36 @@
+import streamlit as st
+import json
+from PIL import Image
+
+
+def load_image(image_path, image_resize=None):
+    image = Image.open(image_path)
+    if isinstance(image_resize, tuple):
+        image.resize(image_resize)
+    return image
+
+
+def load_text(text_path):
+    text = ''
+    with open(text_path) as f:
+        text = f.read()
+
+    return text
+
+
+def load_json(json_path):
+    jdata = ''
+    with open(json_path) as f:
+        jdata = json.load(f)
+
+    return jdata
+
+
+def local_css(css_path):
+    with open(css_path) as f:
+        st.markdown(f'<style>{f.read()}</style>', unsafe_allow_html=True)
+
+
+def remote_css(css_url):
+    st.markdown(f'<link href="{css_url}" rel="stylesheet">', unsafe_allow_html=True)
+