src/pipeline.py

from diffusers import AutoencoderKL, AutoencoderTiny
from diffusers.image_processor import VaeImageProcessor
import torch
import torch._dynamo
import gc
from PIL.Image import Image
from pipelines.models import TextToImageRequest
from torch import Generator
from diffusers import FluxPipeline
from torchao.quantization import quantize_, int8_weight_only, fpx_weight_only
import torch.nn as nn 
from model import Model, Decoder, Encoder
import torchvision

Pipeline = None
MODEL_ID = "black-forest-labs/FLUX.1-schnell"
DTYPE = torch.bfloat16
def clear():
    gc.collect()
    torch.cuda.empty_cache()
    torch.cuda.reset_max_memory_allocated()
    torch.cuda.reset_peak_memory_stats()

def load_pipeline() -> Pipeline:    
    clear()
    
    # vae = Model("encoder.pth", "decoder.pth")
    # vae.to(dtype=DTYPE)

    vae = AutoencoderTiny.from_pretrained("madebyollin/taef1")
    vae.encoder = Encoder(16)
    vae.decoder = Decoder(16)

    encoder_path = "encoder.pth"
    decoder_path = "decoder.pth"

    if encoder_path is not None:
        encoder_state_dict = torch.load(encoder_path, map_location="cpu", weights_only=True)
        filtered_state_dict = {k.strip('encoder.'): v for k, v in encoder_state_dict.items() if k.strip('encoder.') in vae.encoder.state_dict() and v.size() == vae.encoder.state_dict()[k.strip('encoder.')].size()}
        print(f" num of keys in filtered: {len(filtered_state_dict)} and in decoder: {len(vae.encoder.state_dict())}")
        vae.encoder.load_state_dict(filtered_state_dict, strict=False)
            
    if decoder_path is not None:
        decoder_state_dict = torch.load(decoder_path, map_location="cpu", weights_only=True)
        filtered_state_dict = {k.strip('decoder.'): v for k, v in decoder_state_dict.items() if k.strip('decoder.') in vae.decoder.state_dict() and v.size() == vae.decoder.state_dict()[k.strip('decoder.')].size()}
        print(f" num of keys in filtered: {len(filtered_state_dict)} and in decoder: {len(vae.decoder.state_dict())}")
        vae.decoder.load_state_dict(filtered_state_dict, strict=False)

    vae.decoder.requires_grad_(False)
    vae.encoder.requires_grad_(False)
    vae.to(dtype=DTYPE)
    
    pipeline = FluxPipeline.from_pretrained(MODEL_ID,vae=vae,
                                        torch_dtype=DTYPE)
    torch.backends.cudnn.benchmark = True
    torch.backends.cuda.matmul.allow_tf32 = True
    torch.cuda.set_per_process_memory_fraction(0.99)
    pipeline.text_encoder.to(memory_format=torch.channels_last)
    pipeline.text_encoder_2.to(memory_format=torch.channels_last)
    pipeline.transformer.to(memory_format=torch.channels_last)
    pipeline.vae.to(memory_format=torch.channels_last)
    pipeline.vae = torch.compile(pipeline.vae)
    pipeline._exclude_from_cpu_offload = ["vae"]
    pipeline.enable_sequential_cpu_offload()

    clear()
    for _ in range(1):
        pipeline(prompt="unpervaded, unencumber, froggish, groundneedle, transnatural, fatherhood, outjump, cinerator", width=1024, height=1024, guidance_scale=0.1, num_inference_steps=4, max_sequence_length=256)
    return pipeline

sample = True
@torch.inference_mode()
def infer(request: TextToImageRequest, pipeline: Pipeline) -> Image:
    global sample
    if sample:
        clear()
        sample = None
    torch.cuda.reset_peak_memory_stats()
    generator = Generator("cuda").manual_seed(request.seed)
    image = None
    # with torch.backends.cuda.sdp_kernel(enable_flash=True, enable_math=False, enable_mem_efficient=False):
    # image=pipeline(request.prompt,generator=generator, guidance_scale=0.0, num_inference_steps=4, max_sequence_length=256, height=request.height, width=request.width, output_type="pt").images[0]
    image=pipeline(request.prompt,generator=generator, guidance_scale=0.0, num_inference_steps=4, max_sequence_length=256, height=request.height, width=request.width, output_type="pil").images[0]
    # image = image / 255.
    # image = image.mul_(2).sub_(1)
    # image = ((image + 1) / 2) * 255
    # image = image.clamp(0, 255)
    # image = image.to(torch.float32)
    # return torchvision.transforms.functional.to_pil_image(image)
    # return torchvision.transforms.functional.to_pil_image(image.to(torch.float32).mul_(2).sub_(1))
    return image