update README to have less changes to the previous version

README.md

@@ -5,15 +5,14 @@ base_model: Qwen/Qwen-Image-2.1
 library_name: diffusers
 pipeline_tag: text-to-image
 tags:
-- qwen
-- image-generation
-- image-editing
-- rgba
-- lora
-- distilled
-- few-step
+  - qwen
+  - image-generation
+  - image-editing
+  - rgba
+  - lora
+  - distilled
+  - few-step
 ---
-
 <!-- header start -->
 <!-- 200823 -->
 <p align="center">
@@ -42,42 +41,49 @@ tags:
     <img src="https://img.shields.io/badge/Performance%20Models-Try%20them%20now-9334E9?style=plastic&logo=data:image/svg%2bxml;base64,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" alt="Performance Models">
   </a>
 </p>
+<div align="center">
 
-# ⚡ Pruna-Q-Image-2.1
-
-
+<h1 style="color: #9334E9;">⚡ Pruna-Q-Image-2.1</h1>
 
-## Few-step LoRA adapters for Qwen-Image-2.1
+<h2>Few-step LoRA adapters for Qwen-Image-2.1</h2>
 
+<h3>
+  <span style="color: #9334E9;">6 steps (v0.2), 5 or 8 steps (v0.1)</span>
+  &nbsp;·&nbsp;
+  <span style="color: #9334E9;">Up to 6.3x faster</span>
+  &nbsp;·&nbsp;
+  <span style="color: #9334E9;">No CFG</span>
+  &nbsp;·&nbsp;
+  Text-to-image and image editing
+</h3>
 
+</div>
 
-### 6 steps  ·  No CFG  ·  Text-to-image and image editing
-
-**Pruna-Q-Image-2.1 is a set of LoRA adapters that let**
-`[Qwen/Qwen-Image-2.1](https://huggingface.co/Qwen/Qwen-Image-2.1)` **generate
-and edit images in only 6 steps. The adapters load on top of the base
+**Pruna-Q-Image-2.1 is a set of LoRA adapters that let
+[`Qwen/Qwen-Image-2.1`](https://huggingface.co/Qwen/Qwen-Image-2.1) generate
+and edit images in only 6 steps (v0.2), or 5 or 8 steps (v0.1). The adapters load on top of the base
 model, so the pipeline, text encoder, and VAE stay unchanged. Training is based
 on DMD. Improved using Qwen.**
 
 > **v0.2: two new 6-step adapters.** A basic distillation for faithful generation
 > and editing, and a first attempt at distilling prompt upsampling into the LoRA.
-> The original v0.1 adapters remain available.
-
+> The v0.1 adapters remain available.
 
+> **v0.1: first release, work in progress.** Pruna-Q-Image-2.1 does not yet match the
+> visual quality of the base model. We are still improving the distillation and
+> will update this repository as new versions become available.
 
 ## Variants
 
+### v0.2
+
 Pick **basic** for faithful base-model behavior, or **prompt-upsampled** for
 improved visual quality from a short prompt.
 
-
-| Adapter                   | File                                      | Steps | Trade-off                                                                                      |
-| ------------------------- | ----------------------------------------- | ----- | ---------------------------------------------------------------------------------------------- |
-| **Basic v0.2**            | `p_q_image_2.1_basic_6_steps_v0.2.safetensors`                      | 6     | **Pure few-step distillation.** Recommended for preserving inputs when editing.                |
-| **Prompt-upsampled v0.2** | `p_q_image_2.1_upsampled_6_steps_v0.2.safetensors`                      | 6     | **Improved visual quality.** A first attempt at distilling prompt upsampling into the adapter. |
-| v0.1, 8-step              | `p_q_image_2.1_8step_v0.1.safetensors` | 8     | **Higher quality** of the two original adapters.                                               |
-| v0.1, 5-step              | `p_q_image_2.1_5step_v0.1.safetensors` | 5     | **Higher speed**, with noticeably lower visual quality.                                        |
-
+| File | Steps | Trade-off |
+|---|:---:|---|
+| `p_q_image_2.1_basic_6_steps_v0.2.safetensors` | 6 | **Pure few-step distillation.** Recommended for preserving inputs when editing. |
+| `p_q_image_2.1_upsampled_6_steps_v0.2.safetensors` | 6 | **Improved visual quality.** A first attempt at distilling prompt upsampling into the adapter. |
 
 **Basic** is a pure 6-step distillation of the base model, without any additional
 post-training. It is intended as a drop-in replacement for the base model in
@@ -90,11 +96,22 @@ Qwen-Image-2.1 into a 6-step LoRA. It improves visual quality without a separate
 prompt-upsampling call. This is our first attempt: it may not reproduce exactly
 what the actual prompt upsampler would do, and we will continue improving it.
 
-Each adapter is trained for its own sigma schedule. Load only one at a time.  
-Do not reuse the v0.1 schedules for the new adapters.
+### v0.1
+
+Both adapters are v0.1. Pick one based on whether you need **quality** or **speed**.
+
+| File | Steps | Trade-off |
+|---|:---:|---|
+| `p_q_image_2.1_8step_v0.1.safetensors` | 8 | **Higher quality.** Recommended as the default. |
+| `p_q_image_2.1_5step_v0.1.safetensors` | 5 | **Higher speed**, with noticeably lower visual quality. |
+
+Each adapter is trained for its own sigma schedule. Load only one at a time.
+Do not reuse the v0.1 schedules for the v0.2 adapters.
 
 ## Prompts, editing, and resolution
 
+For **v0.2**:
+
 For **basic**, use the same prompts and reference images you would use with the
 base model. For **prompt-upsampled**, start with your original prompt; the adapter
 is intended to capture the benefit of prompt upsampling during generation.
@@ -102,35 +119,33 @@ Detailed prompts still let you specify the subject, setting, lighting, and style
 
 The adapters were trained at **1K resolution only**, with a mix of simple and
 upsampled prompts, text-to-image generation, and single- and multi-image editing
-with **up to 3 reference images**. Start at **1024 × 1024** and use at most
-**3 reference images** with those adapters. Higher resolutions and more reference
-images are outside their training coverage, and quality may vary.
+with **up to 3 reference images**. Prompt upsampling is optional; detailed prompts
+usually give better results.
 
-The 2K timings below show inference speed, not a guarantee of quality at that
-resolution.
+Start at **1024 × 1024** and use at most **3 reference images** for editing.
+Higher resolutions (including 2K) and more reference images may work, but are
+outside the training coverage, and quality may vary. The 2K timings below show
+inference speed, not a guarantee of quality at that resolution.
 
 ## Benchmarking
 
-**v0.1 timings.** These charts show the original 5- and 8-step adapters.
-The 6-step adapter and separate prompt-upsampling measurements are reported
-separately below.
-
+Use a table; Hugging Face can otherwise render the `<img>` elements as separate blocks.
 
 <table>
   <tr>
     <td width="50%">
-      <img src="assets/Pruna-Q-Image-2.1-t2i-1024.png" alt="v0.1 text-to-image latency at 1024 pixels" width="100%">
+      <img src="assets/Pruna-Q-Image-2.1-t2i-1024.png" width="100%">
     </td>
     <td width="50%">
-      <img src="assets/Pruna-Q-Image-2.1-t2i-2048.png" alt="v0.1 text-to-image latency at 2048 pixels" width="100%">
+      <img src="assets/Pruna-Q-Image-2.1-t2i-2048.png" width="100%">
     </td>
   </tr>
   <tr>
     <td width="50%">
-      <img src="assets/Pruna-Q-Image-2.1-i2i-1024.png" alt="v0.1 image-editing latency at 1024 pixels" width="100%">
+      <img src="assets/Pruna-Q-Image-2.1-i2i-1024.png" width="100%">
     </td>
     <td width="50%">
-      <img src="assets/Pruna-Q-Image-2.1-i2i-2048.png" alt="v0.1 image-editing latency at 2048 pixels" width="100%">
+      <img src="assets/Pruna-Q-Image-2.1-i2i-2048.png" width="100%">
     </td>
   </tr>
 </table>
@@ -138,12 +153,12 @@ separately below.
 Text-to-image with the official Qwen model-card example prompt, BF16, batch size 1,
 on one NVIDIA H100 80GB. Median of 3 requests after one warmup per configuration.
 Includes prompt encoding, denoising, and decoding; excludes PNG saving, model
-loading, and warmup. Base: 40 steps with KV cache on; v0.1 adapters: 5 or 8 steps with
-KV cache off, as configured for this benchmark. The examples below keep diffusion KV caching off, matching this setting. LoRAs were unmerged;
+loading, and warmup. Base: 40 steps with KV cache on; Pruna-Q-Image-2.1: 5 or 8 steps with
+KV cache off, as configured for this benchmark. The examples below enable KV caching for
+Pruna-Q-Image-2.1; the chart has not been remeasured with that setting. LoRAs were unmerged;
 no CFG, compilation, or CPU offload. These timings do not imply equal image quality.
 
-### 6-step LoRA and prompt upsampling
-
+### v0.2: 6-step LoRA and prompt upsampling
 
 <table>
   <tr>
@@ -166,15 +181,15 @@ processing, denoising, and VAE decoding to a PIL image. It excludes model and
 adapter loading, warmup, fixture preparation, file saving, and network transport.
 No CFG, compilation, CPU offload, or diffusion KV caching.
 
-Prompt upsampling is measured separately using  
-`[Qwen-Image-2.1-PE-T2I](https://huggingface.co/Qwen/Qwen-Image-2.1-PE-T2I)`  
-and `[Qwen-Image-2.1-PE-I2I](https://huggingface.co/Qwen/Qwen-Image-2.1-PE-I2I)`,  
-with their task-specific sampling settings and thinking enabled. Timing includes  
-request preparation from in-memory images, image and text processing, text  
-generation, and response parsing. It does not include image generation. The  
-upsampler uses vLLM in eager mode, without prefix or image-processor caching.  
-Each request receives fresh image identifiers, so vision encoding is included  
-on repeated requests too. Autoregressive KV caching remains enabled; it is  
+Prompt upsampling is measured separately using
+[`Qwen-Image-2.1-PE-T2I`](https://huggingface.co/Qwen/Qwen-Image-2.1-PE-T2I)
+and [`Qwen-Image-2.1-PE-I2I`](https://huggingface.co/Qwen/Qwen-Image-2.1-PE-I2I),
+with their task-specific sampling settings and thinking enabled. Timing includes
+request preparation from in-memory images, image and text processing, text
+generation, and response parsing. It does not include image generation. The
+upsampler uses vLLM in eager mode, without prefix or image-processor caching.
+Each request receives fresh image identifiers, so vision encoding is included
+on repeated requests too. Autoregressive KV caching remains enabled; it is
 separate from diffusion KV caching.
 
 Each editing case uses a fixed original prompt and 1, 2, or 3 reference images,
@@ -182,7 +197,7 @@ square-cropped and resized to the requested resolution before timing.
 Text-to-image uses the official Qwen model-card example prompt. Latency depends
 on the prompt, image
 content, and generated rewrite length. These cases are not a controlled
-comparison with the older v0.1 chart inputs. Adding the two medians does not
+comparison with the v0.1 chart inputs. Adding the two medians does not
 measure an end-to-end prompt-upsampling-plus-generation request.
 
 Software and prompt-upsampling settings
@@ -191,7 +206,7 @@ Generation uses PyTorch 2.11.0 with CUDA 12.8, Transformers 5.20.1, PEFT 0.21.2,
 and Diffusers commit `6256aa7666cedd47443adc8f82da9a10e110b09c`, the same Diffusers
 revision used in the quickstart. Prompt upsampling uses vLLM 0.19.1, PyTorch
 2.10.0, and Transformers 5.19.0. These runtime versions are recorded for the new
-measurements; the older charts have not been rerun in this environment.
+measurements; the v0.1 charts have not been rerun in this environment.
 
 Both prompt enhancers use their checkpoint's system prompt, thinking enabled,
 `temperature=1.0`, `top_p=0.95`, `top_k=20`, and `min_p=0.0`. Text-to-image uses
@@ -209,6 +224,42 @@ pip install 'torch>=2.4.0' 'transformers>=5.17' accelerate peft pillow
 pip install git+https://github.com/huggingface/diffusers@6256aa7666cedd47443adc8f82da9a10e110b09c
 ```
 
+### v0.1
+
+```python
+import torch
+from PIL import Image
+from diffusers import FlowMatchEulerDiscreteScheduler, QwenImage21Pipeline
+
+STEPS = 8  # 8 for higher quality, 5 for higher speed
+
+# The terminal sigma 0 is appended by the scheduler.
+SIGMAS = {
+    5: [1.0, 0.94, 6 / 7, 2 / 3, 0.4],
+    8: [1.0, 14 / 15, 6 / 7, 10 / 13, 2 / 3, 6 / 11, 0.4, 2 / 9],
+}[STEPS]
+
+pipe = QwenImage21Pipeline.from_pretrained(
+    "Qwen/Qwen-Image-2.1", torch_dtype=torch.bfloat16
+).to("cuda")
+pipe.load_lora_weights(
+    "PrunaAI/Pruna-Q-Image-2.1",
+    weight_name=f"p_q_image_2.1_{STEPS}step_v0.1.safetensors",
+)
+
+# Use the sigmas exactly as given: no extra shifting.
+pipe.scheduler = FlowMatchEulerDiscreteScheduler.from_config(
+    pipe.scheduler.config,
+    use_dynamic_shifting=False,
+    shift=1.0,
+    shift_terminal=None,
+)
+```
+
+### v0.2
+
+Use this setup instead of the v0.1 setup above for either new adapter.
+
 ```python
 import torch
 from PIL import Image
@@ -241,7 +292,6 @@ pipe.scheduler = FlowMatchEulerDiscreteScheduler.from_config(
 ```
 
 
-
 Both variants load on **`Qwen/Qwen-Image-2.1`** using **`QwenImage21Pipeline`**.
 Choose one LoRA with `VARIANT`; the prompt-upsampled variant runs directly on your
 original prompt without a separate prompt-enhancer model. For a local export,
@@ -252,35 +302,12 @@ Both files have **rank 64** across **232 linear targets**, including `proj_out`.
 Training used alpha 128; the exported B matrices include the `128 / 64 = 2`
 scale. Load at **LoRA strength 1.0** with the standard pipeline loader.
 
-### Using the original v0.1 adapters
-
-To use an original adapter, replace the `STEPS`, `WEIGHT_NAME`, and `SIGMAS`
-settings above with the following and reset the scheduler as shown. The v0.1
-sigmas are already shifted, so dynamic shifting must be disabled for them.
-
-```python
-STEPS = 8  # 8 for higher quality, 5 for higher speed
-WEIGHT_NAME = f"p_q_image_2.1_{STEPS}step_v0.1.safetensors"
-SIGMAS = {
-    5: [1.0, 0.94, 6 / 7, 2 / 3, 0.4],
-    8: [1.0, 14 / 15, 6 / 7, 10 / 13, 2 / 3, 6 / 11, 0.4, 2 / 9],
-}[STEPS]
-
-pipe.unload_lora_weights()
-pipe.load_lora_weights("PrunaAI/Pruna-Q-Image-2.1", weight_name=WEIGHT_NAME)
-pipe.scheduler = FlowMatchEulerDiscreteScheduler.from_config(
-    pipe.scheduler.config,
-    use_dynamic_shifting=False,
-    shift=1.0,
-    shift_terminal=None,
-)
-```
-
-
+For v0.2 benchmark settings, pass `output_resolution=1024` and
+`use_kv_cache=False` in the generation and editing calls below.
 
 ## For Text-to-Image
 
-Use a detailed prompt when you want precise control over the result.
+Longer, more descriptive prompts give better results.
 
 ```python
 image = pipe(
@@ -291,73 +318,73 @@ image = pipe(
     ),
     width=1024,
     height=1024,
-    output_resolution=1024,
     generator=torch.Generator("cuda").manual_seed(42),
     num_inference_steps=STEPS,
     sigmas=SIGMAS,
     true_cfg_scale=1.0,
-    use_kv_cache=False,
+    use_kv_cache=True,
 ).images[0]
 ```
 
-
-
 ### For Image Editing
 
 ```python
 image = pipe(
     prompt="Change the background to a sunset beach",
     image=Image.open("input.png").convert("RGB"),
-    output_resolution=1024,
     generator=torch.Generator("cuda").manual_seed(42),
     num_inference_steps=STEPS,
     sigmas=SIGMAS,
     true_cfg_scale=1.0,
-    use_kv_cache=False,
+    use_kv_cache=True,
 ).images[0]
 ```
 
+### Recommended settings
 
+For **v0.2**, use the unshifted six-step sigmas
+`[1.0, 0.925, 0.85, 0.625, 0.375, 0.125]` with dynamic shifting enabled
+and `shift_terminal=None`, as shown in the v0.2 setup. The scheduler appends
+terminal zero. Keep `true_cfg_scale=1.0` and LoRA strength 1.0.
 
-### Recommended settings
+The **v0.1** settings are:
 
-- **Use the sigma schedule and scheduler settings that match your adapter.**
-  - v0.2, both 6-step adapters: pass the **unshifted** sigmas
-    `[1.0, 0.925, 0.85, 0.625, 0.375, 0.125]` with dynamic exponential shifting
-    enabled using the quickstart's shift parameters. Keep `shift_terminal=None`.
-  - v0.1, 8-step: `1 → 14/15 → 6/7 → 10/13 → 2/3 → 6/11 → 0.4 → 2/9 → 0`.
-  - v0.1, 5-step: `1 → 0.94 → 6/7 → 2/3 → 0.4 → 0`.
-    These v0.1 schedules are already shifted; use dynamic shifting off and `shift=1.0`.
-  - The scheduler appends terminal zero for all variants. Pass each schedule
-    without that final zero.
+- **Use the sigma schedule that matches your adapter**, and keep the scheduler
+  at `shift=1.0` with dynamic shifting off so the sigmas are not shifted twice.
+  - 8-step: `1 → 14/15 → 6/7 → 10/13 → 2/3 → 6/11 → 0.4 → 2/9 → 0`
+    (shift 2, computed as `σ = 2t / (1 + t)` on evenly spaced `t`)
+  - 5-step: `1 → 0.94 → 6/7 → 2/3 → 0.4 → 0`
 - **No CFG.** Keep `true_cfg_scale=1.0` and do not pass a negative prompt.
 - **Keep the LoRA strength at 1.0.**
-- **Choose the adapter for your workflow.** Use basic for faithful generation
-and editing, or prompt-upsampled to capture some of the benefit of prompt
-upsampling without running it separately.
-
-
+- **Write detailed prompts for text-to-image.** Longer prompts that describe
+  the subject, setting, lighting, and style work noticeably better than short
+  ones.
 
 ## Limitations
 
+### v0.2
+
 - The prompt-upsampled adapter is a first attempt. It can improve visual quality,
-but may not exactly reproduce the output of the actual prompt upsampler.
+  but may not exactly reproduce the output of the actual prompt upsampler.
 - Basic is intended to preserve the base model's behavior in fewer steps;
-distillation does not guarantee identical outputs.
-- The original v0.1 adapters remain below the base model in visual quality.
-The 5-step adapter is faster than the 8-step one, but its images are visibly worse.
-- Use the step count and sigma schedule that match your adapter. Other step
-counts, schedules, or CFG values are not supported.
+  distillation does not guarantee identical outputs.
+- Use the step count and sigma schedule that match your v0.2 adapter. Other step
+  counts, schedules, or CFG values are not supported.
 
+### v0.1
 
+- This is a first version. Quality is below that of the base model.
+- The 5-step adapter is faster than the 8-step one, but its images are
+  visibly worse.
+- Short or vague text-to-image prompts give weaker results.
+- The adapters are only intended for the settings above. Other step counts,
+  schedules, or CFG values are not supported.
 
 ## What this is not
 
 - Not a standalone model. It needs the `Qwen/Qwen-Image-2.1` base weights.
-- Not an exact reproduction of the separate prompt upsampler.
-- Not a finished release. Expect further improvements in future versions.
-
-
+- Not a replacement for the base model when you need its full quality.
+- Not a finished release. Expect the weights to change in future versions.
 
 ## License
 
@@ -376,4 +403,32 @@ Hangzhou Tongyi Laboratory Technology Co., Ltd. All Rights Reserved.
 - Compress your own models with [Pruna](https://github.com/PrunaAI/pruna) and give us a ⭐️ for more efficiency!
 - Want to use our optimized image models right away?  Check [P-Image-Ideogram](https://www.pruna.ai/p-image-ideogram) and [P-Image-Edit](https://www.pruna.ai/p-image-edit).
 
-[Try our models](https://dashboard.pruna.ai/login?utm_source=huggingface&utm_medium=model_card&utm_campaign=hf_traffic)
+<style>
+.model-button {
+  display: inline-flex;
+  flex-direction: row;
+  justify-content: center;
+  align-items: center;
+  gap: 8px;
+
+  padding: 8px 20px;
+  border: none;
+  border-radius: 8px;
+
+  background: #9334e9;
+  color: #ffffff;
+
+  font-size: 14px;
+  font-weight: 400;
+  line-height: 1;
+  text-decoration: none;
+  white-space: nowrap;
+  cursor: pointer;
+
+  box-sizing: border-box;
+  overflow: visible;
+  opacity: 1;
+}
+</style>
+
+<a href="https://dashboard.pruna.ai/login?utm_source=huggingface&utm_medium=model_card&utm_campaign=hf_traffic" class="model-button">Try our models</a>