import html
import gradio as gr
import modules.textual_inversion.textual_inversion as ti
from modules import sd_hijack, shared
def create_embedding(name, nvpt):
filename = ti.create_embedding(name, nvpt)
sd_hijack.model_hijack.embedding_db.load_textual_inversion_embeddings()
return gr.Dropdown.update(choices=sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())), f"Created: {filename}", ""
def train_embedding(*args):
try:
sd_hijack.undo_optimizations()
embedding, filename = ti.train_embedding(*args)
res = f"""
Training {'interrupted' if shared.state.interrupted else 'finished'} after {embedding.step} steps.
Embedding saved to {html.escape(filename)}
"""
return res, ""
except Exception:
raise
finally:
sd_hijack.apply_optimizations()