Spaces:
Build error
Build error
File size: 3,241 Bytes
8b891df e45afa6 8b891df e45afa6 8b891df 6b8803d e45afa6 8b891df e45afa6 8b891df e45afa6 8b891df 08c842e e45afa6 8b891df e45afa6 |
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 |
import gradio as gr
import torch
import os
from PIL import Image
from pathlib import Path
from more_itertools import chunked
from transformers import CLIPProcessor, CLIPModel
checkpoint = "vincentclaes/emoji-predictor"
x_, _, files = next(os.walk("./emojis"))
no_of_emojis = range(len(files))
emojis_as_images = [Image.open(f"emojis/{i}.png") for i in no_of_emojis]
K = 4
processor = CLIPProcessor.from_pretrained(checkpoint)
model = CLIPModel.from_pretrained(checkpoint)
def concat_images(*images):
"""Generate composite of all supplied images.
https://stackoverflow.com/a/71315656/1771155
"""
# Get the widest width.
width = max(image.width for image in images)
# Add up all the heights.
height = max(image.height for image in images)
# set the correct size of width and heigtht of composite.
composite = Image.new('RGB', (2*width, 2*height))
assert K == 4, "We expect 4 suggestions, other numbers won't work."
for i, image in enumerate(images):
if i == 0:
composite.paste(image, (0, 0))
elif i == 1:
composite.paste(image, (width, 0))
elif i == 2:
composite.paste(image, (0, height))
elif i == 3:
composite.paste(image, (width, height))
return composite
def get_emoji(text, model=model, processor=processor, emojis=emojis_as_images, K=4):
inputs = processor(text=text, images=emojis, return_tensors="pt", padding=True, truncation=True)
outputs = model(**inputs)
logits_per_text = outputs.logits_per_text
# we take the softmax to get the label probabilities
probs = logits_per_text.softmax(dim=1)
# top K number of options
predictions_suggestions_for_chunk = [torch.topk(prob, K).indices.tolist() for prob in probs][0]
predictions_suggestions_for_chunk
images = [Image.open(f"emojis/{i}.png") for i in predictions_suggestions_for_chunk]
images_concat = concat_images(*images)
return images_concat
text = gr.inputs.Textbox(placeholder="Enter a text and we will try to predict an emoji...")
title = "Predicting an Emoji"
description = """You provide a sentence and our few-shot fine tuned CLIP model will predict from the following emoji's:
\nβ€οΈ π π π π₯ π π β¨ π π π· πΊπΈ β π π π― π π πΈ π βΉοΈ π π π‘ π’ π€ π³ π π© π π π\n
"""
article = """
\n
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
\n
We fine tuned Open Ai's CLIP model on both text (tweets) and images of emoji's!\n
The current model is fine-tuned on 15 samples per emoji.
- model: https://huggingface.co/vincentclaes/emoji-predictor \n
- dataset: https://huggingface.co/datasets/vincentclaes/emoji-predictor \n
- code: https://github.com/vincentclaes/emoji-predictor \n
- profile: https://huggingface.co/vincentclaes \n
"""
examples = [
"I'm so happy for you!",
"I'm not feeling great today.",
"This makes me angry!",
"Can I follow you?",
"I'm so bored right now ...",
]
gr.Interface(fn=get_emoji, inputs=text, outputs=gr.Image(shape=(72,72)),
examples=examples, title=title, description=description,
article=article).launch()
|