Upload README.md with huggingface_hub

README.md

@@ -11,9 +11,9 @@ base_model: nvidia/PixelDiT-1300M-1024px
 
 # PixelDiT 1.3B — Diffusers-Compatible Conversion
 
-This is an **unofficial** HuggingFace-compatible conversion of NVIDIA's [PixelDiT-1300M-1024px](https://huggingface.co/nvidia/PixelDiT-1300M-1024px) model.
+This is an **unofficial** HuggingFace diffusers-compatible conversion of NVIDIA's [PixelDiT-1300M-1024px](https://huggingface.co/nvidia/PixelDiT-1300M-1024px).
 
-All credit goes to the original authors at NVIDIA. This repo only provides a `PreTrainedModel` wrapper to enable `from_pretrained`, `save_pretrained`, and LoRA fine-tuning via `peft`.
+All credit goes to the original authors at NVIDIA. This repo only provides a `DiffusionPipeline` wrapper to enable standard diffusers usage, `from_pretrained`, and LoRA fine-tuning via `peft`.
 
 > **I do not own this model.** Original weights, architecture, and training are the work of NVIDIA Research. Please refer to their [original repository](https://huggingface.co/nvidia/PixelDiT-1300M-1024px) for license terms.
 
@@ -21,41 +21,67 @@ All credit goes to the original authors at NVIDIA. This repo only provides a `Pr
 
 ## What is PixelDiT?
 
-PixelDiT is a 1.3B parameter pixel-space diffusion transformer — no VAE, generates images directly in pixel space. Text conditioning uses Gemma-2-2B with a chi_prompt prefix to produce rich visual descriptions.
+PixelDiT is a 1.3B parameter **pixel-space** diffusion transformer — no VAE, generates images directly in pixel space. Runs on **4GB VRAM** at 512px.
 
 - **Architecture**: MMDiT patch blocks + pixel pathway (PiT blocks)
-- **Text encoder**: Gemma-2-2B (`Efficient-Large-Model/gemma-2-2b-it`)
+- **Text encoder**: Gemma-2-2B with chi_prompt instruction prefix
 - **Resolution**: up to 1024×1024
-- **Sampler**: Flow matching (DPM-Solver++ recommended, 20 steps)
+- **Sampler**: Flow matching (FlowMatchEulerDiscreteScheduler, shift=4.0)
 
 ---
 
-## Usage
-
-```python
-from pixeldit import PixelDiTPipeline
-
-pipe = PixelDiTPipeline(pretrained="madtune/pixeldit-diffusers")
-img = pipe("a white horse running in a meadow at sunset", height=512, width=512)[0]
-img.save("out.jpg")
-```
+## Install
 
-Install the package:
 ```bash
-git clone https://github.com/madtune/pixeldit-diffusers
+git clone https://github.com/madtunebk/pixeldit-diffusers
 cd pixeldit-diffusers
+python setup_diffusers_pixeldit.py
 pip install transformers accelerate safetensors pillow
 ```
 
 ---
 
+## Usage
+
+```python
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+from diffusers.pipelines.pixeldit import PixelDiTPipeline
+
+tokenizer = AutoTokenizer.from_pretrained("Efficient-Large-Model/gemma-2-2b-it")
+tokenizer.padding_side = "right"
+text_encoder = (
+    AutoModelForCausalLM.from_pretrained("Efficient-Large-Model/gemma-2-2b-it", torch_dtype=torch.float32)
+    .get_decoder().eval()
+)
+
+pipe = PixelDiTPipeline.from_pretrained(
+    "madtune/pixeldit-diffusers",
+    text_encoder=text_encoder,
+    tokenizer=tokenizer,
+    torch_dtype=torch.bfloat16,
+)
+pipe.enable_model_cpu_offload()
+
+image = pipe(
+    "a white horse galloping through a meadow at sunset, cinematic lighting",
+    negative_prompt="blurry, flat, low quality, cartoon",
+    height=512, width=512,
+    num_inference_steps=20,
+    guidance_scale=3.5,
+).images[0]
+image.save("out.jpg")
+```
+
+---
+
 ## LoRA fine-tuning
 
 ```python
-from pixeldit import PixelDiTModel
 from peft import get_peft_model, LoraConfig
+from diffusers.pipelines.pixeldit import PixelDiTModel
 
-model = PixelDiTModel.from_pretrained("madtune/pixeldit-diffusers")
+model = PixelDiTModel.from_pretrained("madtune/pixeldit-diffusers", subfolder="transformer")
 lora_cfg = LoraConfig(target_modules=["qkv_x", "qkv_y", "proj_x", "proj_y"])
 model = get_peft_model(model, lora_cfg)
 model.print_trainable_parameters()
@@ -63,8 +89,17 @@ model.print_trainable_parameters()
 
 ---
 
+## Sample outputs
+
+| Prompt | Image |
+|--------|-------|
+| *a viking warrior at sunset* | cinematic, photorealistic |
+| *elemental goddess with fire and ice powers* | epic fantasy, 1024px |
+
+---
+
 ## Credits
 
-- **Original model**: [NVIDIA Research](https://huggingface.co/nvidia/PixelDiT-1300M-1024px)
-- **Diffusers conversion**: [madtune](https://huggingface.co/madtune)
+- **Original model & all credit**: [NVIDIA Research](https://huggingface.co/nvidia/PixelDiT-1300M-1024px)
 - **Paper**: *PixelDiT: Pixel-Space Diffusion Transformers for Text-to-Image Generation* — NVIDIA
+- **This repo**: unofficial diffusers conversion only, no claim of authorship