github/InvokeAI
1
1
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI.git synced 2026-04-23 03:00:31 -04:00
Code Issues Packages Projects Releases Wiki Activity

move all prompting stuff to use compel

Browse Source
This commit is contained in:
Damian Stewart
2023-02-19 20:42:29 +01:00
parent b9ecf93ba3
commit ded3f13a33
11 changed files with 104 additions and 1646 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
ldm/invoke/generator/diffusers_pipeline.py
View File

@@ -31,7 +31,7 @@ from ldm.modules.textual_inversion_manager import TextualInversionManager
from ..devices import normalize_device, CPU_DEVICE
from ..offloading import LazilyLoadedModelGroup, FullyLoadedModelGroup, ModelGroup
from ...models.diffusion.cross_attention_map_saving import AttentionMapSaver
from ...modules.prompt_to_embeddings_converter import WeightedPromptFragmentsToEmbeddingsConverter
from compel import EmbeddingsProvider
@dataclass
@@ -294,7 +294,7 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
text_encoder=self.text_encoder,
full_precision=use_full_precision)
# InvokeAI's interface for text embeddings and whatnot
self.prompt_fragments_to_embeddings_converter = WeightedPromptFragmentsToEmbeddingsConverter(
self.embeddings_provider = EmbeddingsProvider(
tokenizer=self.tokenizer,
text_encoder=self.text_encoder,
textual_inversion_manager=self.textual_inversion_manager
@@ -726,15 +726,15 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
"""
Compatibility function for ldm.models.diffusion.ddpm.LatentDiffusion.
"""
return self.prompt_fragments_to_embeddings_converter.get_embeddings_for_weighted_prompt_fragments(
text=c,
fragment_weights=fragment_weights,
return self.embeddings_provider.get_embeddings_for_weighted_prompt_fragments(
text_batch=c,
fragment_weights_batch=fragment_weights,
should_return_tokens=return_tokens,
device=self._model_group.device_for(self.unet))
@property
def cond_stage_model(self):
return self.prompt_fragments_to_embeddings_converter
return self.embeddings_provider
@torch.inference_mode()
def _tokenize(self, prompt: Union[str, List[str]]):