Spaces:
Runtime error
Runtime error
Amir Zait
commited on
Commit
·
be37091
1
Parent(s):
0d9345a
added dalle
Browse files- app.py +10 -3
- image_generator.py +46 -0
- requirements.txt +4 -0
app.py
CHANGED
|
@@ -8,6 +8,8 @@ import torch
|
|
| 8 |
import sox
|
| 9 |
import os
|
| 10 |
|
|
|
|
|
|
|
| 11 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 12 |
|
| 13 |
api_token = os.getenv("API_TOKEN")
|
|
@@ -49,6 +51,9 @@ def convert(inputfile, outfile):
|
|
| 49 |
)
|
| 50 |
sox_tfm.build(inputfile, outfile)
|
| 51 |
|
|
|
|
|
|
|
|
|
|
| 52 |
def parse_transcription(wav_file):
|
| 53 |
filename = wav_file.name.split('.')[0]
|
| 54 |
convert(wav_file.name, filename + "16k.wav")
|
|
@@ -58,10 +63,12 @@ def parse_transcription(wav_file):
|
|
| 58 |
logits = asr_model(input_values).logits
|
| 59 |
predicted_ids = torch.argmax(logits, dim=-1)
|
| 60 |
transcription = asr_processor.decode(predicted_ids[0], skip_special_tokens=True)
|
| 61 |
-
translated = he_en_translator(transcription)
|
| 62 |
-
|
|
|
|
|
|
|
| 63 |
|
| 64 |
-
output = gr.outputs.
|
| 65 |
input_mic = gr.inputs.Audio(source="microphone", type="file", optional=True)
|
| 66 |
input_upload = gr.inputs.Audio(source="upload", type="file", optional=True)
|
| 67 |
|
|
|
|
| 8 |
import sox
|
| 9 |
import os
|
| 10 |
|
| 11 |
+
from image_generator import generate_image
|
| 12 |
+
|
| 13 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 14 |
|
| 15 |
api_token = os.getenv("API_TOKEN")
|
|
|
|
| 51 |
)
|
| 52 |
sox_tfm.build(inputfile, outfile)
|
| 53 |
|
| 54 |
+
def generate_image(text):
|
| 55 |
+
pass
|
| 56 |
+
|
| 57 |
def parse_transcription(wav_file):
|
| 58 |
filename = wav_file.name.split('.')[0]
|
| 59 |
convert(wav_file.name, filename + "16k.wav")
|
|
|
|
| 63 |
logits = asr_model(input_values).logits
|
| 64 |
predicted_ids = torch.argmax(logits, dim=-1)
|
| 65 |
transcription = asr_processor.decode(predicted_ids[0], skip_special_tokens=True)
|
| 66 |
+
translated = he_en_translator(transcription)[0]['translation_text']
|
| 67 |
+
|
| 68 |
+
image = generate_image(translated)
|
| 69 |
+
return image
|
| 70 |
|
| 71 |
+
output = gr.outputs.Image(label='')
|
| 72 |
input_mic = gr.inputs.Audio(source="microphone", type="file", optional=True)
|
| 73 |
input_upload = gr.inputs.Audio(source="upload", type="file", optional=True)
|
| 74 |
|
image_generator.py
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import random
|
| 2 |
+
import numpy as np
|
| 3 |
+
from PIL import Image
|
| 4 |
+
|
| 5 |
+
from dalle_mini import DalleBart, DalleBartProcessor
|
| 6 |
+
from vqgan_jax.modeling_flax_vqgan import VQModel
|
| 7 |
+
|
| 8 |
+
# Model references
|
| 9 |
+
|
| 10 |
+
# dalle-mega
|
| 11 |
+
DALLE_MODEL = "dalle-mini/dalle-mini/mega-1-fp16:latest" # can be wandb artifact or 🤗 Hub or local folder or google bucket
|
| 12 |
+
DALLE_COMMIT_ID = None
|
| 13 |
+
|
| 14 |
+
# if the notebook crashes too often you can use dalle-mini instead by uncommenting below line
|
| 15 |
+
# DALLE_MODEL = "dalle-mini/dalle-mini/mini-1:v0"
|
| 16 |
+
|
| 17 |
+
# VQGAN model
|
| 18 |
+
VQGAN_REPO = "dalle-mini/vqgan_imagenet_f16_16384"
|
| 19 |
+
VQGAN_COMMIT_ID = "e93a26e7707683d349bf5d5c41c5b0ef69b677a9"
|
| 20 |
+
|
| 21 |
+
model = DalleBart.from_pretrained(DALLE_MODEL, revision=DALLE_COMMIT_ID)
|
| 22 |
+
vqgan = VQModel.from_pretrained(VQGAN_REPO, revision=VQGAN_COMMIT_ID)
|
| 23 |
+
processor = DalleBartProcessor.from_pretrained(DALLE_MODEL, revision=DALLE_COMMIT_ID)
|
| 24 |
+
|
| 25 |
+
def get_image(text):
|
| 26 |
+
tokenized_prompt = processor([text])
|
| 27 |
+
|
| 28 |
+
gen_top_k = None
|
| 29 |
+
gen_top_p = None
|
| 30 |
+
temperature = 0.85
|
| 31 |
+
cond_scale = 3.0
|
| 32 |
+
|
| 33 |
+
encoded_images = model.generate(
|
| 34 |
+
tokenized_prompt,
|
| 35 |
+
random.randint(0, 1e7),
|
| 36 |
+
model.params,
|
| 37 |
+
gen_top_k,
|
| 38 |
+
gen_top_p,
|
| 39 |
+
temperature,
|
| 40 |
+
cond_scale,
|
| 41 |
+
)
|
| 42 |
+
encoded_images = encoded_images.sequences[..., 1:]
|
| 43 |
+
decoded_images = model.decode(encoded_images, vqgan.params)
|
| 44 |
+
decoded_images = decoded_images.clip(0.0, 1.0).reshape((-1, 256, 256, 3))
|
| 45 |
+
img = decoded_images[0]
|
| 46 |
+
return Image.fromarray(np.asarray(img * 255, dtype=np.uint8))
|
requirements.txt
CHANGED
|
@@ -5,3 +5,7 @@ torch
|
|
| 5 |
transformers
|
| 6 |
sox
|
| 7 |
sentencepiece
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
transformers
|
| 6 |
sox
|
| 7 |
sentencepiece
|
| 8 |
+
vqgan-jax
|
| 9 |
+
dalle-mini
|
| 10 |
+
PIL
|
| 11 |
+
numpy
|