prodypanda/pulire-tdm-lora-v1

Pulire Tdm Lora V1 - LoRA Concept Adapter

This is a LoRA (Low-Rank Adaptation) adapter trained on the pulire-tdm-lora-v1 concept using the a photo of <pulire-tdm> towel dispenser machine trigger. It was trained on the base model prodypanda/pulire-towel-dispenser-concept-v1.

Trigger Prompt: a photo of <pulire-tdm> towel dispenser machine

Usage (with � Diffusers)

from diffusers import StableDiffusionPipeline, AutoencoderKL
import torch

# 1. Load the base model pipeline
base_model_id = "prodypanda/pulire-towel-dispenser-concept-v1"
# Optional: Load a specific VAE if needed
# vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16)
# pipe = StableDiffusionPipeline.from_pretrained(base_model_id, vae=vae, torch_dtype=torch.float16)
pipe = StableDiffusionPipeline.from_pretrained(base_model_id, torch_dtype=torch.float16)
pipe.to("cuda")

# 2. Load the LoRA adapter weights
lora_adapter_id = "prodypanda/pulire-tdm-lora-v1"
pipe.load_lora_weights(lora_adapter_id)
# Optional: Specify subfolders if weights are organized that way in the repo
# pipe.load_lora_weights(lora_adapter_id, subfolder="unet", weight_name="pytorch_lora_weights.safetensors")
# if text_encoder LoRA exists:
# pipe.load_lora_weights(lora_adapter_id, subfolder="text_encoder", weight_name="pytorch_lora_weights.safetensors")


# 3. Generate images!
prompt = "a photo of <pulire-tdm> towel dispenser machine in a vibrant jungle"
negative_prompt = "low quality, blurry, unrealistic"

# Adjust LoRA weight (optional, 0.0-1.0) - requires Diffusers >= 0.17.0
# image = pipe(prompt, negative_prompt=negative_prompt, cross_attention_kwargs={"scale": 0.8}).images[0]

image = pipe(prompt, negative_prompt=negative_prompt).images[0]
image.save("output_lora.png")

# To unload LoRA and use the base model again:
# pipe.unload_lora_weights()

Training Images

The following images were used for training this concept:

---

LoRA training run using the � Diffusers and � PEFT libraries.