v1.4: icon grounding, not-on-screen refusals, symbol names; Windows +2.5
Browse files- README.md +86 -74
- model.safetensors +1 -1
README.md
CHANGED
|
@@ -6,13 +6,15 @@ tags: [system-one, decision-model, gui-grounding, computer-use, voice-control, t
|
|
| 6 |
base_model: Qwen/Qwen3.5-0.8B-Base
|
| 7 |
---
|
| 8 |
|
| 9 |
-
# OpenThai-SystemOne-Vision (v1.
|
| 10 |
|
| 11 |
The vision variant of [OpenThai-SystemOne](https://huggingface.co/iapp/OpenThai-SystemOne): the same open Thai + English
|
| 12 |
System One decision model, now with **images in the state** and a fourth question type, **`point`**, that returns where
|
| 13 |
on a screenshot something is. Screenshot in, click out, in one forward pass, no text generation.
|
| 14 |
|
| 15 |
-
**v1.
|
|
|
|
|
|
|
| 16 |
large screens, and adds `point_refine`;
|
| 17 |
**v1.1 (2026-09-24)** added Thai **voice-command understanding** for hands-free computer control (a spoken Thai command +
|
| 18 |
the app and window title in, the operation and its target out) and stronger desktop and Doom play. See [Versions](#versions).
|
|
@@ -70,41 +72,43 @@ state = 'ΰΈΰΈ³ΰΈͺΰΈ±ΰΉΰΈΰΉΰΈͺΰΈ΅ΰΈ’ΰΈΰΈ ΰΈ²ΰΈ©ΰΈ²ΰΉΰΈΰΈ’ΰΈΰΈ²ΰΈΰΈΰΈΉ
|
|
| 70 |
|
| 71 |
## Evaluation
|
| 72 |
|
| 73 |
-
v1.
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
### Live desktops
|
| 79 |
-
| benchmark | v1.
|
| 80 |
-
|---|---|---|---|---|---|
|
| 81 |
-
| **Windows live** (319 targets, 28 screens of Windows 11: Settings, File Explorer, Control Panel, Notepad, Paint, Calculator, Store, Edge, β¦) |
|
| 82 |
-
| Β· Windows Settings pages (114) |
|
| 83 |
-
|
|
| 84 |
-
|
|
| 85 |
-
|
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
|
|
|
| 90 |
|
| 91 |
### GUI grounding benchmarks (`point`: predicted click inside the target box)
|
| 92 |
-
| benchmark | v1.
|
| 93 |
-
|---|---|---|---|---|---|
|
| 94 |
-
| ScreenSpot-v2, all (1,272) |
|
| 95 |
-
| Β· desktop text / icon |
|
| 96 |
-
| Β· mobile text / icon | 94.5 /
|
| 97 |
-
| Β· web text / icon |
|
| 98 |
-
| ScreenSpot-Pro (1,581, 26 pro apps
|
| 99 |
-
| Desktop (OS-Atlas macOS / Windows / Linux, held-out) |
|
| 100 |
-
| Desktop "not on screen" abstention (26) |
|
| 101 |
-
| Thai GUI held-out domains, `point`
|
| 102 |
-
| Thai GUI set-of-marks `choice` / `operation` | 93.
|
| 103 |
-
| Thai screen QA `noul` / `choice` | 93.
|
| 104 |
-
| Thai GUI "not on screen" abstention |
|
| 105 |
|
| 106 |
Qwen3-VL numbers: our runs with the same harness (screenshot + instruction, click inside the box); * every 2nd ScreenSpot-Pro
|
| 107 |
-
item (790).
|
|
|
|
| 108 |
|
| 109 |
### Better clicks on large screens: `point_refine`
|
| 110 |
```python
|
|
@@ -114,63 +118,68 @@ r = c.system_one("goal: β¦", {"where": {"type": "point", "instructions": "ΰΈ
|
|
| 114 |
```
|
| 115 |
The first pass points on the whole screen; the client then re-points on a crop around each of the top candidates at native
|
| 116 |
detail and keeps the most confident. It only zooms landscape screens larger than the pixel budget (zooming phone screenshots
|
| 117 |
-
hurts). ScreenSpot-Pro
|
|
|
|
| 118 |
|
| 119 |
### Thai voice commands (operation + slots from one spoken sentence)
|
| 120 |
-
| set | v1.
|
| 121 |
|---|---|---|---|
|
| 122 |
-
| real commands spoken in use, operation (183) |
|
| 123 |
-
| Β· slots (site / app / key / Nth) when the operation is right |
|
| 124 |
-
| Β· chit-chat between commands acted on (41), lower is better |
|
| 125 |
-
| real off-command conversation picked up by the microphone, acted on (227), lower is better |
|
| 126 |
-
| synthetic voice commands, held out (1,406 questions) |
|
| 127 |
-
| synthetic hard cases (fragments, site vs app, ordinals, 446 questions) |
|
| 128 |
-
|
| 129 |
-
### Doom from pixels (`demo/doom_demo.py`, defend_the_center, skill 3, 40 episodes
|
| 130 |
-
|
|
| 131 |
|---|---|---|---|---|
|
| 132 |
-
|
|
|
|
|
|
|
|
| 133 |
| the privileged scripted expert it imitates | 6.8 | 6.8 | 6.8 | β |
|
| 134 |
|
| 135 |
-
|
|
|
|
|
|
|
| 136 |
|
| 137 |
### Text-only tasks
|
| 138 |
Public 13-subset benchmark (the Bespoke / Jev public suite; `noul`/`choice` = accuracy, `score` = exact level):
|
| 139 |
|
| 140 |
-
| subset | v0.3 | v1.
|
| 141 |
|---|---|---|---|---|---|
|
| 142 |
-
| aegis2 | 83.2 | 82.0 |
|
| 143 |
-
| boolq | 79.7 | 79.7 | 79.
|
| 144 |
-
| civil_comments | 79.0 | 86.
|
| 145 |
-
| helpsteer2 (score) | 41.6 |
|
| 146 |
-
| massive-de-DE | 88.3 | 89.
|
| 147 |
-
| massive-en-US | 88.3 |
|
| 148 |
-
| multinli | 89.0 |
|
| 149 |
| paws | 94.0 | 94.0 | 94.0 | 89.2 | 82.8 |
|
| 150 |
-
| pubmedqa | 64.0 | 67.
|
| 151 |
-
| squad2 | 89.3 |
|
| 152 |
-
| summeval-consistency (score) | 75.0 | 75.0 | 75.
|
| 153 |
-
| summeval-relevance (score) | 21.7 |
|
| 154 |
-
| vitaminc-dev | 72.5 | 71.
|
| 155 |
-
| **macro, 13 subsets** | 74.3 |
|
| 156 |
-
| macro, 10 accuracy subsets | 82.7 | 83.7 | 83.
|
| 157 |
|
| 158 |
\* published by Bespoke Labs for their own harness on the same subsets (Jev via the TypeSafe API).
|
| 159 |
|
| 160 |
Thai held-out and eval-split sets (`th` unless noted; banking77 and xLAM are English):
|
| 161 |
|
| 162 |
-
| set | v0.3 | v1.
|
| 163 |
|---|---|---|---|---|
|
| 164 |
-
| banking77 (77-way, whole dataset held out) | 45.4 | 62.0 | 62.
|
| 165 |
-
| contrastive pairs, one-fact flips | 80.7 | 82.1 |
|
| 166 |
-
| MASSIVE-th intent | 90.0 | 91.1 | 91.
|
| 167 |
-
| Prachathai topics | 98.1 | 98.
|
| 168 |
-
| SIB-200 Thai topic (whole dataset held out) | 77.9 | 75.5 | 75.
|
| 169 |
-
| Wisesight sentiment (whole dataset held out) | 51.6 | 52.
|
| 170 |
-
| Wongnai stars (score, exact) | 63.5 | 63.
|
| 171 |
-
| xLAM tool selection | 99.4 | 99.
|
| 172 |
-
| XNLI-th | 77.1 | 75.
|
| 173 |
-
| **macro, 8 accuracy sets** | 77.5 | 79.6 | 79.
|
| 174 |
|
| 175 |
### Latency (batch 1, H100, end-to-end `SystemOneClient`)
|
| 176 |
| request | tokens | ms |
|
|
@@ -184,7 +193,8 @@ Thai held-out and eval-split sets (`th` unless noted; banking77 and xLAM are Eng
|
|
| 184 |
|
| 185 |
| version | date | change |
|
| 186 |
|---|---|---|
|
| 187 |
-
| **v1.
|
|
|
|
| 188 |
| v1.1 | 2026-09-24 | + Thai voice-command intents (28k synthetic training records from 21k distinct commands + 8.9k hard-case records: fragments / chit-chat = NONE, site vs app, ordinal + topic, ASR spelling variants), desktop grounding round (OS-Atlas macOS / Windows / Linux), text-anchor round (civil_comments, VitaminC); temperatures recalibrated. Voice 65 β 88.5, desktop 39 β 51, Doom pixels 5.5 β 11.9 kills, ScreenSpot-v2 82.7 β 80.7. |
|
| 189 |
| v1.0 | 2026-09-23 | first vision release (Thai GUI crawl, public grounding, Doom rollouts, screen QA on the v0.3 text tower) |
|
| 190 |
|
|
@@ -206,7 +216,9 @@ Thai held-out and eval-split sets (`th` unless noted; banking77 and xLAM are Eng
|
|
| 206 |
6. v1.3: one round from v1.2 with live Windows 11 and macOS screens captured through the accessibility APIs (a small
|
| 207 |
agent inside a Windows VM; the macOS accessibility tree), other apps / pages than the live eval screens, personal names
|
| 208 |
blacked out, rest of the mix rehearsed.
|
| 209 |
-
7.
|
|
|
|
|
|
|
| 210 |
|
| 211 |
## License & credits
|
| 212 |
|
|
|
|
| 6 |
base_model: Qwen/Qwen3.5-0.8B-Base
|
| 7 |
---
|
| 8 |
|
| 9 |
+
# OpenThai-SystemOne-Vision (v1.4)
|
| 10 |
|
| 11 |
The vision variant of [OpenThai-SystemOne](https://huggingface.co/iapp/OpenThai-SystemOne): the same open Thai + English
|
| 12 |
System One decision model, now with **images in the state** and a fourth question type, **`point`**, that returns where
|
| 13 |
on a screenshot something is. Screenshot in, click out, in one forward pass, no text generation.
|
| 14 |
|
| 15 |
+
**v1.4 (2026-09-28)** clicks better on Windows apps (Settings, Calculator) and on icons, and understands voice commands a little
|
| 16 |
+
better; it plays Doom from pixels worse than v1.3 (see Evaluation).
|
| 17 |
+
**v1.3 (2026-09-26)** clicked better on real Windows and Mac apps (trained on live Windows 11 and macOS screens) and on
|
| 18 |
large screens, and adds `point_refine`;
|
| 19 |
**v1.1 (2026-09-24)** added Thai **voice-command understanding** for hands-free computer control (a spoken Thai command +
|
| 20 |
the app and window title in, the operation and its target out) and stronger desktop and Doom play. See [Versions](#versions).
|
|
|
|
| 72 |
|
| 73 |
## Evaluation
|
| 74 |
|
| 75 |
+
v1.4 next to v1.3, same scripts; single option order unless noted; accuracy in %. "default" = the plain API call (screenshot at
|
| 76 |
+
β€ 1280Γ800); "+ refine" = screenshot at 2 MP (`"detail": "high"`) with `point_refine` over the top-3 first-pass candidates,
|
| 77 |
+
landscape screens only (zoom 50 % on ScreenSpot-v2 / Thai GUI / OS-Atlas, 35 % on Windows / Mac live and ScreenSpot-Pro).
|
| 78 |
+
On desktops and large screens, send `"detail": "high"` and `point_refine`.
|
| 79 |
+
|
| 80 |
+
### Live desktops: real apps, click targets from the accessibility tree
|
| 81 |
+
| benchmark | v1.3 default | v1.3 + refine | **v1.4 default** | **v1.4 + refine** | Qwen3-VL-2B | Qwen3-VL-4B |
|
| 82 |
+
|---|---|---|---|---|---|---|
|
| 83 |
+
| **Windows live** (319 targets, 28 screens of Windows 11: Settings, File Explorer, Control Panel, Notepad, Paint, Calculator, Store, Edge, β¦) | 68.3 | 69.3 | **70.8** | 70.5 | 67.1 | 73.0 |
|
| 84 |
+
| Β· Windows Settings pages (114) | 69.3 | β | **74.6** | β | 66.7 | 68.4 |
|
| 85 |
+
| Β· Windows Calculator (20)β | 60 | β | **80** | β | 80 | 95 |
|
| 86 |
+
| **Mac live** (358 targets, 29 screens: System Settings, Finder, Safari, Office, iWork, β¦) | 63.7 | 67.6 | 64.5 | 67.3 | 67.0 | 67.9 |
|
| 87 |
+
| Β· label-checked (340: 18 targets found unanswerable removed) | 66.8 | **71.2** | 67.6 | 70.9 | 70.6 | 71.5 |
|
| 88 |
+
| latency per click (H100, eval harness) | ~70 ms | 0.1β0.25 s | ~70 ms | 0.1β0.25 s | ~0.9 s | ~0.75 s |
|
| 89 |
+
|
| 90 |
+
The Windows and Mac live screens are never trained on; the training data includes *other* screens of the same apps.
|
| 91 |
+
β v1.4's synthetic calculators use Windows' standard accessible names ("Two", "Multiply by"), which is how the Windows live
|
| 92 |
+
Calculator targets are named, so part of that gain is learning the naming convention.
|
| 93 |
|
| 94 |
### GUI grounding benchmarks (`point`: predicted click inside the target box)
|
| 95 |
+
| benchmark | v1.3 default | v1.3 + refine | **v1.4 default** | **v1.4 + refine** | Qwen3-VL-2B | Qwen3-VL-4B |
|
| 96 |
+
|---|---|---|---|---|---|---|
|
| 97 |
+
| ScreenSpot-v2, all (1,272) | 82.2 | 85.2 | **82.8** | **85.9** | 85.8 | 90.4 |
|
| 98 |
+
| Β· desktop text / icon | 90.2 / 75.7 | 91.8 / 83.6 | 89.2 / 77.1 | 91.2 / 85.0 | 93.8 / 82.9 | 96.9 / 85.7 |
|
| 99 |
+
| Β· mobile text / icon | 94.5 / 71.6 | 93.8 / 74.9 | 94.8 / 71.6 | 94.5 / 75.4 | 94.1 / 82.0 | 96.9 / 83.4 |
|
| 100 |
+
| Β· web text / icon | 84.2 / 70.0 | 88.5 / 74.9 | 85.9 / 71.4 | 89.3 / 76.4 | 85.9 / 72.4 | 91.9 / 83.7 |
|
| 101 |
+
| ScreenSpot-Pro (1,581, 26 pro apps up to 4K; Windows / macOS / Linux) | 9.0 | 32.9 (29.3 / 38.7 / 28.0) | 8.7 | 32.8 (29.2 / 38.9 / 26.0) | 43.4* | 54.8* |
|
| 102 |
+
| Desktop (OS-Atlas macOS / Windows / Linux, held-out) | 53.7 | 59.8 (600) | 53.8 | 59.3 (600) | 63.8 | 65.6 |
|
| 103 |
+
| Desktop "not on screen" abstention (26) | 92.3 | β | 92.3 | β | β | β |
|
| 104 |
+
| Thai GUI held-out domains, `point` (1,611) | 77.8 | 76.7 (600) | **78.6** | 74.9 (600) | 72.4 | 80.4 |
|
| 105 |
+
| Thai GUI set-of-marks `choice` / `operation` | 93.1 / 99.6 | β | 92.6 / 99.8 | β | β | β |
|
| 106 |
+
| Thai screen QA `noul` / `choice` | 93.1 / 85.9 | β | 94.7 / 85.0 | β | β | β |
|
| 107 |
+
| Thai GUI "not on screen" abstention (179) | 48.0 | β | **49.7** | β | β | β |
|
| 108 |
|
| 109 |
Qwen3-VL numbers: our runs with the same harness (screenshot + instruction, click inside the box); * every 2nd ScreenSpot-Pro
|
| 110 |
+
item (790). Mac live and the 600-item Thai GUI / OS-Atlas subsets move by Β±1.5 points between settings; smaller differences
|
| 111 |
+
are noise.
|
| 112 |
|
| 113 |
### Better clicks on large screens: `point_refine`
|
| 114 |
```python
|
|
|
|
| 118 |
```
|
| 119 |
The first pass points on the whole screen; the client then re-points on a crop around each of the top candidates at native
|
| 120 |
detail and keeps the most confident. It only zooms landscape screens larger than the pixel budget (zooming phone screenshots
|
| 121 |
+
hurts). v1.4: ScreenSpot-Pro 8.7 β 32.8, ScreenSpot-v2 82.8 β 85.9, Mac live 67.6 β 70.9 (checked). Windows windows usually fit
|
| 122 |
+
the 2 MP budget, so there it mostly adds the higher resolution. Cost: 1 + `cands` forward passes.
|
| 123 |
|
| 124 |
### Thai voice commands (operation + slots from one spoken sentence)
|
| 125 |
+
| set | v1.1 | v1.3 | **v1.4** |
|
| 126 |
|---|---|---|---|
|
| 127 |
+
| real commands spoken in use, operation (183) | 88.5 | 89.6 | **90.7** |
|
| 128 |
+
| Β· slots (site / app / key / Nth) when the operation is right | 100.0 | 100.0 | 100.0 |
|
| 129 |
+
| Β· chit-chat between commands acted on (41), lower is better | 7.3 | 7.3 | 7.3 |
|
| 130 |
+
| real off-command conversation picked up by the microphone, acted on (227), lower is better | 2.6 | 4.4 | 4.0 |
|
| 131 |
+
| synthetic voice commands, held out (1,406 questions) | 90.5 | 92.0 | 91.8 |
|
| 132 |
+
| synthetic hard cases (fragments, site vs app, ordinals, 446 questions) | 90.8 | 93.3 | 92.6 |
|
| 133 |
+
|
| 134 |
+
### Doom from pixels (`demo/doom_demo.py`, defend_the_center, skill 3, two 320Γ240 frames, 40 episodes per seed)
|
| 135 |
+
| seed | v1.1 | v1.3 | **v1.4** | ms / decision |
|
| 136 |
|---|---|---|---|---|
|
| 137 |
+
| 999 | 11.8 | 13.0 | 11.4 | ~50 |
|
| 138 |
+
| 777 | β | 11.9 | 9.7 | ~50 |
|
| 139 |
+
| mean of both (80 episodes) | β | **12.5** | 10.6 | |
|
| 140 |
| the privileged scripted expert it imitates | 6.8 | 6.8 | 6.8 | β |
|
| 141 |
|
| 142 |
+
**Known regression:** v1.4 plays Doom from pixels worse than v1.3 (β1.9 kills per episode). Use v1.3 (`revision="v1.3"`) for the
|
| 143 |
+
Doom demo. Offline Doom decisions (600 frames), v1.4: move 88.5, attack 97.2, in-danger 96.7, nearest-enemy point 48.9,
|
| 144 |
+
enemy-absent 98.3.
|
| 145 |
|
| 146 |
### Text-only tasks
|
| 147 |
Public 13-subset benchmark (the Bespoke / Jev public suite; `noul`/`choice` = accuracy, `score` = exact level):
|
| 148 |
|
| 149 |
+
| subset | v0.3 | v1.3 | **v1.4** | Jev 1.13.0* | Nimble-9B* |
|
| 150 |
|---|---|---|---|---|---|
|
| 151 |
+
| aegis2 | 83.2 | 82.0 | 81.2 | 80.4 | 81.2 |
|
| 152 |
+
| boolq | 79.7 | 79.7 | 79.0 | 89.7 | 86.0 |
|
| 153 |
+
| civil_comments | 79.0 | 86.3 | 86.3 | 81.0 | 70.3 |
|
| 154 |
+
| helpsteer2 (score) | 41.6 | 40.4 | 40.4 | 34.1 | 39.0 |
|
| 155 |
+
| massive-de-DE | 88.3 | 89.7 | 90.0 | 86.9 | 83.4 |
|
| 156 |
+
| massive-en-US | 88.3 | 90.3 | 90.3 | 87.4 | 86.9 |
|
| 157 |
+
| multinli | 89.0 | 88.3 | 89.0 | 82.9 | 85.3 |
|
| 158 |
| paws | 94.0 | 94.0 | 94.0 | 89.2 | 82.8 |
|
| 159 |
+
| pubmedqa | 64.0 | 67.2 | 67.2 | 77.2 | 75.6 |
|
| 160 |
+
| squad2 | 89.3 | 88.6 | 88.3 | 82.9 | 80.6 |
|
| 161 |
+
| summeval-consistency (score) | 75.0 | 75.0 | 75.7 | 81.2 | 75.7 |
|
| 162 |
+
| summeval-relevance (score) | 21.7 | 20.4 | 20.4 | 35.0 | 49.2 |
|
| 163 |
+
| vitaminc-dev | 72.5 | 71.0 | 70.8 | 80.1 | 76.6 |
|
| 164 |
+
| **macro, 13 subsets** | 74.3 | 74.8 | 74.8 | 76.0 | 74.8 |
|
| 165 |
+
| macro, 10 accuracy subsets | 82.7 | 83.7 | 83.6 | 83.8 | 80.9 |
|
| 166 |
|
| 167 |
\* published by Bespoke Labs for their own harness on the same subsets (Jev via the TypeSafe API).
|
| 168 |
|
| 169 |
Thai held-out and eval-split sets (`th` unless noted; banking77 and xLAM are English):
|
| 170 |
|
| 171 |
+
| set | v0.3 | v1.3 | **v1.4** | v1.4, 8 orders |
|
| 172 |
|---|---|---|---|---|
|
| 173 |
+
| banking77 (77-way, whole dataset held out) | 45.4 | 62.0 | 62.1 | 64.4 |
|
| 174 |
+
| contrastive pairs, one-fact flips | 80.7 | 82.1 | 81.8 | 80.1 |
|
| 175 |
+
| MASSIVE-th intent | 90.0 | 91.1 | 91.2 | 92.0 |
|
| 176 |
+
| Prachathai topics | 98.1 | 98.6 | 98.6 | 98.7 |
|
| 177 |
+
| SIB-200 Thai topic (whole dataset held out) | 77.9 | 75.5 | 75.0 | 77.0 |
|
| 178 |
+
| Wisesight sentiment (whole dataset held out) | 51.6 | 52.5 | 52.2 | 51.7 |
|
| 179 |
+
| Wongnai stars (score, exact) | 63.5 | 63.3 | 62.9 | 62.9 |
|
| 180 |
+
| xLAM tool selection | 99.4 | 99.4 | 99.4 | 99.3 |
|
| 181 |
+
| XNLI-th | 77.1 | 75.5 | 75.6 | 75.7 |
|
| 182 |
+
| **macro, 8 accuracy sets** | 77.5 | 79.6 | 79.5 | 79.9 |
|
| 183 |
|
| 184 |
### Latency (batch 1, H100, end-to-end `SystemOneClient`)
|
| 185 |
| request | tokens | ms |
|
|
|
|
| 193 |
|
| 194 |
| version | date | change |
|
| 195 |
|---|---|---|
|
| 196 |
+
| **v1.4** | 2026-09-28 | + icon grounding and "not on screen" refusals from xlangai/Jedi (20k icon snaps, 20k refusal rows), 12k synthetic calculators / title bars / toolbars / media controls labelled with their standard accessible names (English + Thai), Mac and Thai GUI rehearsal up; from v1.3. Windows live 68.3 β 70.8 (Settings 69.3 β 74.6, Calculator 60 β 80), ScreenSpot-v2 82.2 β 82.8 (85.9 with refine), Thai not-on-screen 48.0 β 49.7, voice 89.6 β 90.7; **Doom from pixels 12.5 β 10.6 kills**; text, Thai, Mac live, ScreenSpot-Pro, OS-Atlas unchanged within noise. |
|
| 197 |
+
| v1.3 | 2026-09-26 | + live desktop grounding: 665 records from 319 live Windows 11 screens (UI Automation targets; Settings, Control Panel, dialogs, shell folders, Calculator modes, Store, Edge) and 386 records from 116 live macOS screens (accessibility targets; System Settings, Finder, utilities, Safari, Word / Excel / PowerPoint / Pages ribbons), with account names blacked out; on top of an internal v1.2 round (ShowUI desktop + web, 40k zoomed crops, 75k xlangai/Jedi records, 2 MP training, averaged with v1.1). Client / API `point_refine`. Windows live 63.3 β 68.3, ScreenSpot-v2 80.7 β 82.2 (85.2 with refine), ScreenSpot-Pro ~10 β 32.9 with refine, OS-Atlas desktop 50.6 β 53.7, Doom 11.8 β 13.0, text and Thai unchanged. |
|
| 198 |
| v1.1 | 2026-09-24 | + Thai voice-command intents (28k synthetic training records from 21k distinct commands + 8.9k hard-case records: fragments / chit-chat = NONE, site vs app, ordinal + topic, ASR spelling variants), desktop grounding round (OS-Atlas macOS / Windows / Linux), text-anchor round (civil_comments, VitaminC); temperatures recalibrated. Voice 65 β 88.5, desktop 39 β 51, Doom pixels 5.5 β 11.9 kills, ScreenSpot-v2 82.7 β 80.7. |
|
| 199 |
| v1.0 | 2026-09-23 | first vision release (Thai GUI crawl, public grounding, Doom rollouts, screen QA on the v0.3 text tower) |
|
| 200 |
|
|
|
|
| 216 |
6. v1.3: one round from v1.2 with live Windows 11 and macOS screens captured through the accessibility APIs (a small
|
| 217 |
agent inside a Windows VM; the macOS accessibility tree), other apps / pages than the live eval screens, personal names
|
| 218 |
blacked out, rest of the mix rehearsed.
|
| 219 |
+
7. v1.4: one round from v1.3 with icon grounding / refusal data from xlangai/Jedi and synthetic symbol screens, with the Mac and
|
| 220 |
+
Thai GUI data weighted up so they do not slip.
|
| 221 |
+
8. Calibration of the per-type temperatures.
|
| 222 |
|
| 223 |
## License & credits
|
| 224 |
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1724708176
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c46884e5006cc0133ff1959377984f5f4a335900d20d2b318eeedac0864418bb
|
| 3 |
size 1724708176
|