skaissali commited on
Commit
8e955b4
·
verified ·
1 Parent(s): f5913be

Delete README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -171
README.md DELETED
@@ -1,171 +0,0 @@
1
- ---
2
- base_model:
3
- - Qwen/Qwen3-14B
4
- tags:
5
- - text-generation-inference
6
- - transformers
7
- - qwen3
8
- - ui-generation
9
- - peft
10
- - lora
11
- - tailwind-css
12
- - html
13
- language:
14
- - en
15
- ---
16
-
17
- # UIGEN-T3 — Advanced UI Generation with Hybrid Reasoning
18
-
19
- > Tesslate’s next-gen UI model, built for thoughtful design.
20
-
21
- ---
22
-
23
- <img src="https://cdn-uploads.huggingface.co/production/uploads/64d1129297ca59bcf7458d07/fqZE717ubrgSpAOHeMEeQ.png" alt="UIGEN-T3 UI Screenshot 1" width="500">
24
- <img src="https://cdn-uploads.huggingface.co/production/uploads/64d1129297ca59bcf7458d07/N5gfKRD-j9JAeMoGy7O3K.png" alt="UIGEN-T3 UI Screenshot 2" width="500">
25
- <img src="https://cdn-uploads.huggingface.co/production/uploads/64d1129297ca59bcf7458d07/PDGXycDkVsxFRxqHc6TKD.png" alt="UIGEN-T3 UI Screenshot 2" width="500">
26
- <img src="https://cdn-uploads.huggingface.co/production/uploads/64d1129297ca59bcf7458d07/Q4CHYYcDAN60NFjIPVQnH.png" alt="UIGEN-T3 UI Screenshot 3" width="500">
27
- <img src="https://cdn-uploads.huggingface.co/production/uploads/64d1129297ca59bcf7458d07/ePZoCA8lS4Y1KdkchrTlQ.png" alt="UIGEN-T3 UI Screenshot 4" width="500">
28
- <img src="https://cdn-uploads.huggingface.co/production/uploads/64d1129297ca59bcf7458d07/BU-Pk2moNsTOGuYF1XuRx.png" alt="UIGEN-T3 UI Screenshot 5" width="500">
29
- <img src="https://cdn-uploads.huggingface.co/production/uploads/64d1129297ca59bcf7458d07/gz5ub0lDpS-hhw1MB7e07.png" alt="UIGEN-T3 UI Screenshot 6" width="500">
30
-
31
- ## Demos
32
-
33
- Explore New UI generations:
34
- 📂 [https://uigenoutput.tesslate.com](https://uigenoutput.tesslate.com)
35
-
36
- ---
37
-
38
- **Join our Discord:** [https://discord.gg/GNbWAeJ4](https://discord.gg/GNbWAeJ4)
39
- **Our Website:** [https://tesslate.com](https://tesslate.com)
40
-
41
- ## Quick Information
42
-
43
- * **UI generation model** built on **Qwen3 architecture**
44
- * Supports both **components** and **full web pages**
45
- * **Hybrid reasoning system**: Use `/think` or `/no_think` modes
46
- * Powered by **UIGenEval**, a first-of-its-kind benchmark for UI generation
47
- * Released for **research, non-commercial use.** If you want to use it commercially, please contact us for a pilot program.
48
-
49
- ---
50
-
51
- ## Model Details
52
-
53
- * **Base Model**: Qwen/Qwen3-14B
54
- * **Reasoning Style**: Hybrid (`/think` and `/no_think`)
55
- * **Tokenizer**: Qwen default, with design token headers
56
- * **Output**: Components + Full pages (with `<html>`, `<head>`)
57
- * **Images**: User-supplied or placehold.co – no images in the dataset due to licensing concerns.
58
- * **License**: Research only (non-commercial). Contact us for enterprise use cases.
59
-
60
- ---
61
-
62
- ## Reasoning System
63
-
64
- UIGEN-T3 was trained using a **pre/post reasoning model architecture**.
65
-
66
- You can explicitly control the reasoning mode:
67
-
68
- * `/think` → Enables guided reasoning with layout analysis and heuristics.
69
- * `/no_think` → Faster, raw code generation.
70
-
71
- Outputs also include **design tokens** at the top of each generation for easier site-wide customization.
72
-
73
- ---
74
-
75
- ## Inference Parameters
76
-
77
- Please use 20k context length to get the best results if using reasoning.
78
-
79
- | Parameter | Value |
80
- | :---------- | :---- |
81
- | Temperature | 0.6 |
82
- | Top P | 0.95 |
83
- | Top K | 20 |
84
- | Max Tokens | 40k+ |
85
-
86
- ---
87
-
88
- ## Evaluation: UIGenEval Framework
89
-
90
- **UIGenEval** is our internal evaluation suite, designed to **bridge the gap** between creative output and quality assurance.
91
- *(Learn more in our upcoming paper: "UIGenEval: Bridging the Evaluation Gap in AI-Driven UI Generation" - August, 2025)*
92
-
93
- UIGenEval evaluates models across four pillars:
94
-
95
- 1. **Technical Quality** — Clean HTML, CSS structure, semantic accuracy.
96
- 2. **Prompt Adherence** — Feature completeness and fidelity to instructions.
97
- 3. **Interaction Behavior** — Dynamic logic hooks and functional interactivity.
98
- 4. **Responsive Design** — Multi-viewport performance via Lighthouse, Axe-core, and custom scripts.
99
-
100
- This comprehensive framework directly informs our **GRPO reward functions** for the next release.
101
-
102
- ---
103
-
104
- ## Example Prompts to Try
105
-
106
- * `make a google drive clone`
107
- * `build a figma-style canvas with toolbar`
108
- * `create a modern pricing page with three plans`
109
- * `generate a mobile-first recipe sharing app layout`
110
-
111
- ---
112
-
113
- ## Use Cases
114
-
115
- | Use Case | Description |
116
- | :---------------------- | :-------------------------------------------------------------------------- |
117
- | **Startup MVPs** | Quickly scaffold UIs from scratch with clean code. |
118
- | **Design-to-Code Transfer** | Figma (coming soon) → Code generation. |
119
- | **Component Libraries** | Build buttons, cards, navbars, and export at scale. |
120
- | **Internal Tool Builders** | Create admin panels, dashboards, and layout templates. |
121
- | **Rapid Client Prototypes** | Save time on mockups with production-ready HTML+Tailwind outputs. |
122
-
123
- ---
124
-
125
- ## Limitations
126
-
127
- * No Bootstrap support (planned).
128
- * Not suited for production use — **research-only license**.
129
- * Responsive tuning varies across output complexity.
130
-
131
- ---
132
-
133
- ## Roadmap
134
-
135
- | Milestone | Status |
136
- | :-------------------------- | :----- |
137
- | Launch Tesslate Designer | 2 days |
138
- | Figma convert | |
139
- | Bootstrap & JS logic | |
140
- | GRPO fine-tuning | |
141
- | 4B draft model release | Now |
142
-
143
- ---
144
-
145
- ## Technical Requirements
146
-
147
- * **GPU**: ≥16GB VRAM for 14B inference on GGUF.
148
- * **Libraries**: `transformers`, `torch`, `peft`.
149
- * Compatible with Hugging Face inference APIs and local generation pipelines.
150
-
151
- ---
152
-
153
- ## Community & Contribution
154
-
155
- * **Join our Discord:** [https://discord.gg/GNbWAeJ4](https://discord.gg/GNbWAeJ4)
156
- * Chat about AI, design, or model training.
157
- * Want to contribute UIs or feedback? Let’s talk!
158
-
159
- ---
160
-
161
- ## Citation
162
-
163
- ```bibtex
164
- @misc{tesslate_UIGEN-T3,
165
- title={UIGEN-T3: Hybrid Reasoning for Robust UI Generation on Qwen3},
166
- author={Tesslate Team},
167
- year={2025},
168
- publisher={Tesslate},
169
- note={Non-commercial Research License},
170
- url={https://huggingface.co/tesslate/UIGEN-T3}
171
- }