Upload folder using huggingface_hub

.gitattributes

@@ -175,3 +175,14 @@ docs/assets/showcases/interleave/case_0007_bowie_slide_design.webp filter=lfs di
 docs/assets/teaser_2.png filter=lfs diff=lfs merge=lfs -text
 docs/assets/showcases/t2i_infographic/0028.webp filter=lfs diff=lfs merge=lfs -text
 docs/assets/showcases/t2i_infographic/u1-case2.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/perform_vs_speed_avg3.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/perform_vs_speed_avg8.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/showcases/interleave/reasoning.png filter=lfs diff=lfs merge=lfs -text
+docs/assets/showcases/t2i_infographic/0029.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/showcases/t2i_infographic/0030.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/showcases/t2i_infographic/0031.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/showcases/t2i_infographic/0032.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/showcases/t2i_infographic/0033.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/teaser.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/teaser_1.webp filter=lfs diff=lfs merge=lfs -text
+docs/assets/teaser_2.webp filter=lfs diff=lfs merge=lfs -text

docs/inference_infra.md

@@ -78,8 +78,8 @@ see [`deployment.md`](./deployment.md).
 
 ### Generation Performance
 
-The table below is the benchmark template for **2048x2048** image generation.
-Fill in measured numbers for each machine and deployment profile.
+The table below reports **2048x2048** image generation latency for
+**SenseNova-U1-8B-MoT(NEO-Unify)**. Fill in measured numbers for each machine and deployment profile.
 Note: TP2+CFG2 means Tensor Parallel=2 + CFG Parallel=2.
 
 <div align="center">
@@ -100,7 +100,7 @@ In NEO-Unify, the KV cache for the generation stage is provided by the understan
 
 The table below compares the latency of a single diffusion step for
 **2048x2048** image generation with **CFG enabled**. Unless otherwise noted,
-all measurements are taken on **H100**; the `NEO-Unify (TP2+CFG2)` result uses
+all measurements are taken on **H100**; the `TP2+CFG2` result uses
 `2x H100`.
 Note: TP2+CFG2 means Tensor Parallel=2 + CFG Parallel=2.
 
@@ -113,7 +113,7 @@ Note: TP2+CFG2 means Tensor Parallel=2 + CFG Parallel=2.
 | GLM-Image         |      9B       |     7B     |           1.394           |
 | ERNIE-Image       |      8B       |     8B     |           1.565           |
 | LongCat-Image     |      8B       |     6B     |           0.796           |
-| NEO-Unify (1x, no TP/CFG parallelism) | 8B | 8B | 0.312 |
-| NEO-Unify (TP2+CFG2) | 8B | 8B | 0.158 |
+| SenseNova-U1-8B-MoT (Neo-Unify) | 8B | 8B | 0.312 |
+| SenseNova-U1-8B-MoT (Neo-Unify, TP2+CFG2) | 8B | 8B | 0.158 |
 
 </div>

docs/inference_infra_CN.md

@@ -76,7 +76,8 @@ Docker 镜像、启动命令与 API 测试的简明操作手册，请参见 [`de
 
 ### 生成性能
 
-下表为 **2048x2048** 图像生成的基准模板，列出了不同机型与部署配置下的实测数据。
+下表给出 **SenseNova-U1-8B-MoT(NEO-Unify)** 在
+**2048x2048** 图像生成任务上的基准模版。列出了不同机型与部署配置下的实测数据。
 注：TP2+CFG2 表示张量并行=2 + CFG 并行=2。
 
 <div align="center">
@@ -94,7 +95,7 @@ Docker 镜像、启动命令与 API 测试的简明操作手册，请参见 [`de
 
 ### 跨模型速度对比
 
-下表对比了在启用**CFG**条件下，生成 **2048x2048** 图像时单个 diffusion step 的延迟。除特别说明外，所有数据均在 **H100** 上测得；其中 `NEO-Unify (TP2+CFG2)` 使用的是 `2x H100`。
+下表对比了在启用**CFG**条件下，生成 **2048x2048** 图像时单个 diffusion step 的延迟。除特别说明外，所有数据均在 **H100** 上测得；其中 `SenseNova-U1-8B-MoT (NEO-Unify, TP2+CFG2)` 使用的是 `2x H100`。
 注：TP2+CFG2 表示张量并行=2 + CFG 并行=2。
 
 <div align="center">
@@ -106,7 +107,7 @@ Docker 镜像、启动命令与 API 测试的简明操作手册，请参见 [`de
 | GLM-Image                 |    9B    |    7B    |       1.394       |
 | ERNIE-Image               |    8B    |    8B    |       1.565       |
 | LongCat-Image             |    8B    |    6B    |       0.796       |
-| NEO-Unify (1x，无TP/CFG并行) |   8B    |    8B    |       0.312       |
-| NEO-Unify (TP2+CFG2) |    8B    |    8B    |       0.158       |
+| SenseNova-U1-8B-MoT (NEO-Unify) |   8B    |    8B    |       0.312       |
+| SenseNova-U1-8B-MoT (NEO-Unify, TP2+CFG2) |    8B    |    8B    |       0.158       |
 
 </div>

docs/prompt_enhancement.md

@@ -27,8 +27,6 @@ Skip `--enhance` when:
 user prompt ──► LLM (system prompt = infographic expander) ──► expanded prompt ──► SenseNova-U1
 ```
 
-Upstream system prompt: [SenseNova-Skills / u1-infographic](https://github.com/OpenSenseNova/SenseNova-Skills/blob/main/skills/u1-infographic/references/prompts-expand-system.md).
-
 ## 3. Configuration
 
 All configuration is environment-variable based so the same script can
@@ -45,12 +43,23 @@ First, create a `.env` file and populate it with the four required parameters. T
 Add `--print_enhance` to echo the original + enhanced prompt for
 debugging.
 
+To use **SenseNova 6.7 Flash-Lite** as the enhancer, get an API key from
+[SenseNova Console · token-plan](https://platform.sensenova.cn/token-plan),
+then set:
+
+```bash
+U1_ENHANCE_BACKEND=chat_completions
+U1_ENHANCE_ENDPOINT=https://token.sensenova.cn/v1/chat/completions
+U1_ENHANCE_MODEL=sensenova-6.7-flash-lite
+U1_ENHANCE_API_KEY=<your SenseNova API key>
+```
+
 ### 3.1 Recommended backends
 
 | Model | Backend | Endpoint template | Notes |
 | :---- | :------ | :---------------- | :---- |
 | **Gemini 3.1 Pro** (Default) | `chat_completions` | `https://generativelanguage.googleapis.com/v1beta/openai/chat/completions` | Best overall infographic quality in our internal bench. Excellent at structured / hierarchical content. |
-| SenseNova Agentic model | `chat_completions` | _(will be released soon)_ | Comparable to Gemini 3.1 Pro on zh content, cheaper per-token, preferred for production. |
+| SenseNova 6.7 Flash-Lite | `chat_completions` | `https://token.sensenova.cn/v1/chat/completions` | Near Gemini 3.1 Pro quality on Chinese content at lower per-token cost, preferred for production. |
 | Anthropic Claude (Sonnet/Opus) | `anthropic`        | `https://api.anthropic.com/v1/messages` | Strong typography discipline, slightly less "information-dense" out of the box. |
 | Kimi 2.5                      | `chat_completions` | `https://api.moonshot.cn/v1/chat/completions` | Good Chinese enhancements, weaker for English-dense infographics in our runs. |
 | Gemini 3.1 Flash-Lite (Third-party service) | `chat_completions` | `https://aigateway.edgecloudapp.com/v1/f194fd69361cd590f1fa136c9c90eca1/senseai` | The overall quality of the information chart is high and its generation speed is fast. |

docs/showcases.md

@@ -115,6 +115,12 @@ Reproducible prompts are in
     <td align="center"><a href="./assets/showcases/t2i_infographic/0027.webp"><img width="230" alt="t2i image 0024" src="./assets/showcases/t2i_infographic/0027.webp"></a></td>
     <td align="center"><a href="./assets/showcases/t2i_infographic/0026.webp"><img width="230" alt="t2i image 0025" src="./assets/showcases/t2i_infographic/0026.webp"></a></td>
   </tr>
+  <tr>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0029.webp"><img width="230" alt="t2i image 0022" src="./assets/showcases/t2i_infographic/0029.webp"></a></td>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0030.webp"><img width="230" alt="t2i image 0023" src="./assets/showcases/t2i_infographic/0030.webp"></a></td>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0031.webp"><img width="230" alt="t2i image 0024" src="./assets/showcases/t2i_infographic/0031.webp"></a></td>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0032.webp"><img width="230" alt="t2i image 0025" src="./assets/showcases/t2i_infographic/0032.webp"></a></td>
+  </tr>
 </table>
 
 
@@ -235,7 +241,7 @@ Reproducible prompts are in
 
 | |
 | :---: |
-| [<img alt="interleave case 05" src="./assets/showcases/interleave/reasoning_case2.png">](./assets/showcases/interleave/reasoning_case2.png) |
+| [<img alt="interleave case 05" src="./assets/showcases/interleave/reasoning.png">](./assets/showcases/interleave/reasoning.png) |
 
 
 ---

docs/showcases_CN.md

@@ -109,6 +109,12 @@
     <td align="center"><a href="./assets/showcases/t2i_infographic/0027.webp"><img width="230" alt="t2i image 0024" src="./assets/showcases/t2i_infographic/0027.webp"></a></td>
     <td align="center"><a href="./assets/showcases/t2i_infographic/0026.webp"><img width="230" alt="t2i image 0025" src="./assets/showcases/t2i_infographic/0026.webp"></a></td>
   </tr>
+  <tr>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0029.webp"><img width="230" alt="t2i image 0022" src="./assets/showcases/t2i_infographic/0029.webp"></a></td>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0030.webp"><img width="230" alt="t2i image 0023" src="./assets/showcases/t2i_infographic/0030.webp"></a></td>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0031.webp"><img width="230" alt="t2i image 0024" src="./assets/showcases/t2i_infographic/0031.webp"></a></td>
+    <td align="center"><a href="./assets/showcases/t2i_infographic/0032.webp"><img width="230" alt="t2i image 0025" src="./assets/showcases/t2i_infographic/0032.webp"></a></td>
+  </tr>
 </table>
 
 
@@ -214,7 +220,7 @@
 
 | |
 | :---: |
-| [<img alt="interleave reasoning case 2" src="./assets/showcases/interleave/reasoning_case2.png">](./assets/showcases/interleave/reasoning_case2.png) |
+| [<img alt="interleave reasoning case 2" src="./assets/showcases/interleave/reasoning.png">](./assets/showcases/interleave/reasoning.png) |
 
 ---