Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -16,6 +16,10 @@ if not read_token:
|
|
16 |
from huggingface_hub import login
|
17 |
login(read_token)
|
18 |
|
|
|
|
|
|
|
|
|
19 |
# Define a dictionary of conversational models
|
20 |
conversational_models = {
|
21 |
"Qwen": "Qwen/QwQ-32B",
|
@@ -49,12 +53,12 @@ text_to_image_pipelines = {}
|
|
49 |
text_to_speech_pipelines = {}
|
50 |
|
51 |
# Initialize pipelines for other tasks
|
52 |
-
visual_qa_pipeline = pipeline("visual-question-answering", model="dandelin/vilt-b32-finetuned-vqa")
|
53 |
-
document_qa_pipeline = pipeline("question-answering", model="deepset/roberta-base-squad2")
|
54 |
-
image_classification_pipeline = pipeline("image-classification", model="facebook/deit-base-distilled-patch16-224")
|
55 |
-
object_detection_pipeline = pipeline("object-detection", model="facebook/detr-resnet-50")
|
56 |
-
video_classification_pipeline = pipeline("video-classification", model="facebook/timesformer-base-finetuned-k400")
|
57 |
-
summarization_pipeline = pipeline("summarization", model="facebook/bart-large-cnn")
|
58 |
|
59 |
# Load speaker embeddings for text-to-audio
|
60 |
def load_speaker_embeddings(model_name):
|
@@ -62,26 +66,26 @@ def load_speaker_embeddings(model_name):
|
|
62 |
logger.info("Loading speaker embeddings for SpeechT5")
|
63 |
from datasets import load_dataset
|
64 |
dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
|
65 |
-
speaker_embeddings = torch.tensor(dataset[7306]["xvector"]).unsqueeze(0) # Example speaker
|
66 |
return speaker_embeddings
|
67 |
return None
|
68 |
|
69 |
# Use a different model for text-to-audio if stabilityai/stable-audio-open-1.0 is not supported
|
70 |
try:
|
71 |
-
text_to_audio_pipeline = pipeline("text-to-audio", model="stabilityai/stable-audio-open-1.0", use_auth_token=read_token)
|
72 |
except ValueError as e:
|
73 |
logger.error(f"Error loading stabilityai/stable-audio-open-1.0: {e}")
|
74 |
logger.info("Falling back to a different text-to-audio model.")
|
75 |
-
text_to_audio_pipeline = pipeline("text-to-audio", model="microsoft/speecht5_tts")
|
76 |
speaker_embeddings = load_speaker_embeddings("microsoft/speecht5_tts")
|
77 |
|
78 |
-
audio_classification_pipeline = pipeline("audio-classification", model="facebook/wav2vec2-base")
|
79 |
|
80 |
def load_conversational_model(model_name):
|
81 |
if model_name not in conversational_models_loaded:
|
82 |
logger.info(f"Loading conversational model: {model_name}")
|
83 |
tokenizer = AutoTokenizer.from_pretrained(conversational_models[model_name], use_auth_token=read_token)
|
84 |
-
model = AutoModelForCausalLM.from_pretrained(conversational_models[model_name], use_auth_token=read_token)
|
85 |
conversational_tokenizers[model_name] = tokenizer
|
86 |
conversational_models_loaded[model_name] = model
|
87 |
return conversational_tokenizers[model_name], conversational_models_loaded[model_name]
|
@@ -90,7 +94,7 @@ def chat(model_name, user_input, history=[]):
|
|
90 |
tokenizer, model = load_conversational_model(model_name)
|
91 |
|
92 |
# Encode the input
|
93 |
-
input_ids = tokenizer.encode(user_input + tokenizer.eos_token, return_tensors="pt")
|
94 |
|
95 |
# Generate a response
|
96 |
with torch.no_grad():
|
@@ -110,7 +114,7 @@ def generate_image(model_name, prompt):
|
|
110 |
if model_name not in text_to_image_pipelines:
|
111 |
logger.info(f"Loading text-to-image model: {model_name}")
|
112 |
text_to_image_pipelines[model_name] = StableDiffusionPipeline.from_pretrained(
|
113 |
-
text_to_image_models[model_name], use_auth_token=read_token
|
114 |
)
|
115 |
pipeline = text_to_image_pipelines[model_name]
|
116 |
image = pipeline(prompt).images[0]
|
@@ -120,12 +124,14 @@ def generate_speech(model_name, text):
|
|
120 |
if model_name not in text_to_speech_pipelines:
|
121 |
logger.info(f"Loading text-to-speech model: {model_name}")
|
122 |
text_to_speech_pipelines[model_name] = pipeline(
|
123 |
-
"text-to-speech", model=text_to_speech_models[model_name], use_auth_token=read_token
|
124 |
)
|
125 |
pipeline = text_to_speech_pipelines[model_name]
|
126 |
-
audio = pipeline(text)
|
127 |
return audio["audio"]
|
128 |
|
|
|
|
|
129 |
def visual_qa(image, question):
|
130 |
result = visual_qa_pipeline(image, question)
|
131 |
return result["answer"]
|
|
|
16 |
from huggingface_hub import login
|
17 |
login(read_token)
|
18 |
|
19 |
+
# Set device to GPU if available
|
20 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
21 |
+
logger.info(f"Device set to use {device}")
|
22 |
+
|
23 |
# Define a dictionary of conversational models
|
24 |
conversational_models = {
|
25 |
"Qwen": "Qwen/QwQ-32B",
|
|
|
53 |
text_to_speech_pipelines = {}
|
54 |
|
55 |
# Initialize pipelines for other tasks
|
56 |
+
visual_qa_pipeline = pipeline("visual-question-answering", model="dandelin/vilt-b32-finetuned-vqa", device=device)
|
57 |
+
document_qa_pipeline = pipeline("question-answering", model="deepset/roberta-base-squad2", device=device)
|
58 |
+
image_classification_pipeline = pipeline("image-classification", model="facebook/deit-base-distilled-patch16-224", device=device)
|
59 |
+
object_detection_pipeline = pipeline("object-detection", model="facebook/detr-resnet-50", device=device)
|
60 |
+
video_classification_pipeline = pipeline("video-classification", model="facebook/timesformer-base-finetuned-k400", device=device)
|
61 |
+
summarization_pipeline = pipeline("summarization", model="facebook/bart-large-cnn", device=device)
|
62 |
|
63 |
# Load speaker embeddings for text-to-audio
|
64 |
def load_speaker_embeddings(model_name):
|
|
|
66 |
logger.info("Loading speaker embeddings for SpeechT5")
|
67 |
from datasets import load_dataset
|
68 |
dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
|
69 |
+
speaker_embeddings = torch.tensor(dataset[7306]["xvector"]).unsqueeze(0).to(device) # Example speaker
|
70 |
return speaker_embeddings
|
71 |
return None
|
72 |
|
73 |
# Use a different model for text-to-audio if stabilityai/stable-audio-open-1.0 is not supported
|
74 |
try:
|
75 |
+
text_to_audio_pipeline = pipeline("text-to-audio", model="stabilityai/stable-audio-open-1.0", use_auth_token=read_token, device=device)
|
76 |
except ValueError as e:
|
77 |
logger.error(f"Error loading stabilityai/stable-audio-open-1.0: {e}")
|
78 |
logger.info("Falling back to a different text-to-audio model.")
|
79 |
+
text_to_audio_pipeline = pipeline("text-to-audio", model="microsoft/speecht5_tts", use_auth_token=read_token, device=device)
|
80 |
speaker_embeddings = load_speaker_embeddings("microsoft/speecht5_tts")
|
81 |
|
82 |
+
audio_classification_pipeline = pipeline("audio-classification", model="facebook/wav2vec2-base", device=device)
|
83 |
|
84 |
def load_conversational_model(model_name):
|
85 |
if model_name not in conversational_models_loaded:
|
86 |
logger.info(f"Loading conversational model: {model_name}")
|
87 |
tokenizer = AutoTokenizer.from_pretrained(conversational_models[model_name], use_auth_token=read_token)
|
88 |
+
model = AutoModelForCausalLM.from_pretrained(conversational_models[model_name], use_auth_token=read_token).to(device)
|
89 |
conversational_tokenizers[model_name] = tokenizer
|
90 |
conversational_models_loaded[model_name] = model
|
91 |
return conversational_tokenizers[model_name], conversational_models_loaded[model_name]
|
|
|
94 |
tokenizer, model = load_conversational_model(model_name)
|
95 |
|
96 |
# Encode the input
|
97 |
+
input_ids = tokenizer.encode(user_input + tokenizer.eos_token, return_tensors="pt").to(device)
|
98 |
|
99 |
# Generate a response
|
100 |
with torch.no_grad():
|
|
|
114 |
if model_name not in text_to_image_pipelines:
|
115 |
logger.info(f"Loading text-to-image model: {model_name}")
|
116 |
text_to_image_pipelines[model_name] = StableDiffusionPipeline.from_pretrained(
|
117 |
+
text_to_image_models[model_name], use_auth_token=read_token, torch_dtype=torch.float16, device_map="auto"
|
118 |
)
|
119 |
pipeline = text_to_image_pipelines[model_name]
|
120 |
image = pipeline(prompt).images[0]
|
|
|
124 |
if model_name not in text_to_speech_pipelines:
|
125 |
logger.info(f"Loading text-to-speech model: {model_name}")
|
126 |
text_to_speech_pipelines[model_name] = pipeline(
|
127 |
+
"text-to-speech", model=text_to_speech_models[model_name], use_auth_token=read_token, device=device
|
128 |
)
|
129 |
pipeline = text_to_speech_pipelines[model_name]
|
130 |
+
audio = pipeline(text, speaker_embeddings=speaker_embeddings)
|
131 |
return audio["audio"]
|
132 |
|
133 |
+
|
134 |
+
|
135 |
def visual_qa(image, question):
|
136 |
result = visual_qa_pipeline(image, question)
|
137 |
return result["answer"]
|