Upload README.md
Browse files
README.md
CHANGED
|
@@ -1,31 +1,3 @@
|
|
| 1 |
-
---
|
| 2 |
-
configs:
|
| 3 |
-
- config_name: specs
|
| 4 |
-
default: true
|
| 5 |
-
data_files:
|
| 6 |
-
- split: full
|
| 7 |
-
path: viewer_data/specs.parquet
|
| 8 |
-
- config_name: checklists
|
| 9 |
-
data_files:
|
| 10 |
-
- split: full
|
| 11 |
-
path: viewer_data/checklists.parquet
|
| 12 |
-
- config_name: qa_items
|
| 13 |
-
data_files:
|
| 14 |
-
- split: full
|
| 15 |
-
path: viewer_data/qa_items.parquet
|
| 16 |
-
- config_name: video_sources
|
| 17 |
-
data_files:
|
| 18 |
-
- split: full
|
| 19 |
-
path: viewer_data/video_sources.parquet
|
| 20 |
-
- config_name: source_annotations
|
| 21 |
-
data_files:
|
| 22 |
-
- split: full
|
| 23 |
-
path: viewer_data/source_annotations.parquet
|
| 24 |
-
- config_name: rules
|
| 25 |
-
data_files:
|
| 26 |
-
- split: full
|
| 27 |
-
path: viewer_data/rules.parquet
|
| 28 |
-
---
|
| 29 |
<!-- markdownlint-disable MD013 MD033 MD041 MD060 -->
|
| 30 |
|
| 31 |
<p align="center">
|
|
@@ -36,7 +8,7 @@ configs:
|
|
| 36 |
|
| 37 |
<div align="center">
|
| 38 |
|
| 39 |
-
π [Homepage](https://csu-jpg.github.io/LabInstruct.github.io/) | π [Dataset](#data-files) | π Paper
|
| 40 |
|
| 41 |
</div>
|
| 42 |
|
|
@@ -73,8 +45,12 @@ configs:
|
|
| 73 |
|
| 74 |
## π Abstract
|
| 75 |
|
|
|
|
|
|
|
| 76 |
Self-driving laboratories (SDLs) aim to automate the full experimental loop, from scientific decision-making to physical execution. Ideally, AI-generated plans could be carried out directly by robotic systems, but reliable automation remains difficult in complex, open-world laboratory environments, where experiments often involve fine-grained manipulation, long-horizon procedures, and substantial variation across tasks and setups. Humans therefore remain an important execution interface between AI-generated plans and physical experiments, creating a need for clear and effective human-facing experimental guidance. Because laboratory procedures are inherently visual, spatial, and dynamic, video is particularly well suited to communicating apparatus configurations, manipulation actions, temporal dependencies, and state changes. Recent advances in video generation now make it possible to synthesize experimental demonstrations directly from an initial workspace image and a natural-language instruction. However, whether such models can reliably communicate real laboratory procedures has not been systematically studied. We introduce **LabInstruct**, a benchmark for situated instructional video generation in real laboratories. LabInstruct contains 204 tasks across 5 scientific disciplines, with real reference executions and structured annotations of objects, actions, contacts, and state transitions. Evaluating 8 frontier image-to-video models, we find that visually plausible generations frequently remain procedurally incorrect, revealing a substantial gap between visual realism and the reliability required for experimental instruction.
|
| 77 |
|
|
|
|
|
|
|
| 78 |
<a id="project-overview"></a>
|
| 79 |
|
| 80 |
## π Project Overview
|
|
@@ -228,18 +204,21 @@ These prompts are written to produce structured drafts for human verification.
|
|
| 228 |
|
| 229 |
## π Leaderboard
|
| 230 |
|
| 231 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 232 |
|
| 233 |
-
|
| 234 |
-
|
| 235 |
-
| **1** | **MiniMax H3** | Open weight | MiniMax | **46.3** |
|
| 236 |
-
| 2 | Seedance 2.0 β | Commercial | ByteDance | 45.1 |
|
| 237 |
-
| 3 | Wan 3.0 β | Commercial | Alibaba | 44.4 |
|
| 238 |
-
| 4 | Wan 2.2-I2V-A14B | Open weight | Alibaba | 25.9 |
|
| 239 |
-
| 5 | Cosmos3 Super | Open weight | NVIDIA | 22.8 |
|
| 240 |
-
| 6 | LTX 2.3 | Open weight | Lightricks | 22.1 |
|
| 241 |
-
| 7 | LingBot Video | Open weight | Robbyant | 19.5 |
|
| 242 |
-
| 8 | Cosmos3 Nano | Open weight | NVIDIA | 19.4 |
|
| 243 |
|
| 244 |
<a id="bibtex"></a>
|
| 245 |
|
|
@@ -252,10 +231,11 @@ If you find our work helpful, please consider citing it:
|
|
| 252 |
title = {LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures},
|
| 253 |
author = {Yuming Fu and Weijia Wu and Jing Chen and Jiahao Tang and Feifei Chen and Hongyu Zhu and Xin Jin and Alex Jinpeng Wang},
|
| 254 |
year = {2026},
|
| 255 |
-
eprint = {
|
| 256 |
archivePrefix = {arXiv},
|
| 257 |
-
primaryClass = {cs.
|
| 258 |
-
url = {https://arxiv.org/abs/
|
|
|
|
| 259 |
```
|
| 260 |
|
| 261 |
<a id="contact"></a>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
<!-- markdownlint-disable MD013 MD033 MD041 MD060 -->
|
| 2 |
|
| 3 |
<p align="center">
|
|
|
|
| 8 |
|
| 9 |
<div align="center">
|
| 10 |
|
| 11 |
+
π [Homepage](https://csu-jpg.github.io/LabInstruct.github.io/) | π [Dataset](#data-files) | π [Paper](https://arxiv.org/abs/2610.14020) | π» [Code](https://github.com/CSU-JPG/LabInstruct) | π [Leaderboard](#leaderboard)
|
| 12 |
|
| 13 |
</div>
|
| 14 |
|
|
|
|
| 45 |
|
| 46 |
## π Abstract
|
| 47 |
|
| 48 |
+
<div align="justify">
|
| 49 |
+
|
| 50 |
Self-driving laboratories (SDLs) aim to automate the full experimental loop, from scientific decision-making to physical execution. Ideally, AI-generated plans could be carried out directly by robotic systems, but reliable automation remains difficult in complex, open-world laboratory environments, where experiments often involve fine-grained manipulation, long-horizon procedures, and substantial variation across tasks and setups. Humans therefore remain an important execution interface between AI-generated plans and physical experiments, creating a need for clear and effective human-facing experimental guidance. Because laboratory procedures are inherently visual, spatial, and dynamic, video is particularly well suited to communicating apparatus configurations, manipulation actions, temporal dependencies, and state changes. Recent advances in video generation now make it possible to synthesize experimental demonstrations directly from an initial workspace image and a natural-language instruction. However, whether such models can reliably communicate real laboratory procedures has not been systematically studied. We introduce **LabInstruct**, a benchmark for situated instructional video generation in real laboratories. LabInstruct contains 204 tasks across 5 scientific disciplines, with real reference executions and structured annotations of objects, actions, contacts, and state transitions. Evaluating 8 frontier image-to-video models, we find that visually plausible generations frequently remain procedurally incorrect, revealing a substantial gap between visual realism and the reliability required for experimental instruction.
|
| 51 |
|
| 52 |
+
</div>
|
| 53 |
+
|
| 54 |
<a id="project-overview"></a>
|
| 55 |
|
| 56 |
## π Project Overview
|
|
|
|
| 204 |
|
| 205 |
## π Leaderboard
|
| 206 |
|
| 207 |
+
GPT-5.6 Sol is evaluated at 4 FPS; Human is the pooled overall human rating. Higher is better; β marks commercial models.
|
| 208 |
+
|
| 209 |
+
| Rank | Model | Type | Organization | GPT-5.6 Sol | Human | Tier |
|
| 210 |
+
| :--: | :-- | :-- | :-- | --: | --: | :--: |
|
| 211 |
+
| **1** | **MiniMax H3** | Open weight | MiniMax | **46.3** | **44.3** | **T0** |
|
| 212 |
+
| 2 | Seedance 2.0 β | Commercial | ByteDance | 45.1 | 43.0 | T0 |
|
| 213 |
+
| 3 | Wan 3.0 β | Commercial | Alibaba | 44.4 | 37.3 | T0 |
|
| 214 |
+
| 4 | Wan 2.2-I2V-A14B | Open weight | Alibaba | 25.9 | 15.4 | T1 |
|
| 215 |
+
| 5 | Cosmos3 Super | Open weight | NVIDIA | 22.8 | 11.5 | T1 |
|
| 216 |
+
| 6 | LTX 2.3 | Open weight | Lightricks | 22.1 | 3.0 | T2 |
|
| 217 |
+
| 7 | LingBot Video | Open weight | Robbyant | 19.5 | 10.5 | T2 |
|
| 218 |
+
| 8 | Cosmos3 Nano | Open weight | NVIDIA | 19.4 | 5.8 | T2 |
|
| 219 |
|
| 220 |
+
> [!NOTE]
|
| 221 |
+
> Model rankings shift slightly across evaluation views, evaluators, sampling rates, and generation seeds, so we also report a Tier to keep the comparison robust.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 222 |
|
| 223 |
<a id="bibtex"></a>
|
| 224 |
|
|
|
|
| 231 |
title = {LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures},
|
| 232 |
author = {Yuming Fu and Weijia Wu and Jing Chen and Jiahao Tang and Feifei Chen and Hongyu Zhu and Xin Jin and Alex Jinpeng Wang},
|
| 233 |
year = {2026},
|
| 234 |
+
eprint = {2610.14020},
|
| 235 |
archivePrefix = {arXiv},
|
| 236 |
+
primaryClass = {cs.RO},
|
| 237 |
+
url = {https://arxiv.org/abs/2610.14020}
|
| 238 |
+
}
|
| 239 |
```
|
| 240 |
|
| 241 |
<a id="contact"></a>
|