MCPcopy Create free account
hub / github.com/ModelTC/LightX2V / load_models

Method load_models

lightx2v/disagg/services/encoder.py:422–441  ·  view source on GitHub ↗
(self)

Source from the content-addressed store, hash-verified

420 self.data_sender[data_bootstrap_room] = DataSender(self.data_mgr, data_bootstrap_addr, data_bootstrap_room)
421
422 def load_models(self):
423 self.logger.info("Loading Encoder Models...")
424
425 # T5 Text Encoder
426 text_encoders = load_wan_text_encoder(self.config)
427 self.text_encoder = text_encoders[0] if text_encoders else None
428
429 # CLIP Image Encoder (Optional per usage in wan_i2v.py)
430 if self.config.get("use_image_encoder", False):
431 self.image_encoder = load_wan_image_encoder(self.config)
432
433 # VAE Encoder (Required for I2V)
434 # Note: wan_i2v.py logic: if vae_encoder is None: raise RuntimeError
435 # But we only load if needed or always? Let's check the config flags.
436 # It seems always loaded for I2V task, but might be offloaded.
437 # For simplicity of this service, we load it if the task implies it or just try to load.
438 # But `load_wan_vae_encoder` will look at the config.
439 self.vae_encoder = load_wan_vae_encoder(self.config)
440
441 self.logger.info("Encoder Models loaded successfully.")
442
443 def _get_latent_shape_with_lat_hw(self, latent_h, latent_w):
444 return [

Callers 1

__init__Method · 0.95

Calls 5

load_wan_text_encoderFunction · 0.90
load_wan_image_encoderFunction · 0.90
load_wan_vae_encoderFunction · 0.90
infoMethod · 0.80
getMethod · 0.45

Tested by

no test coverage detected