d3-ultra: Space and free API links, video Box and Track
Browse files- MODEL_MANIFEST.json +3 -3
- README.md +3 -0
MODEL_MANIFEST.json
CHANGED
|
@@ -594,11 +594,11 @@
|
|
| 594 |
}
|
| 595 |
]
|
| 596 |
},
|
| 597 |
-
"built_utc": "2026-10-
|
| 598 |
"files_sha256": {
|
| 599 |
"LICENSE": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4",
|
| 600 |
"NOTICE": "d7522881de6b8417b997953c9b82d54e1be61952571a163c2fcf77f1e199406d",
|
| 601 |
-
"README.md": "
|
| 602 |
"assets/CREDITS.md": "43efdd9d3932e86a57839489cb0acd41cc03f41f5b0480bf58ef23ee8b78331c",
|
| 603 |
"assets/ask-once.png": "de25de9c5a99aa1bf675d920839a98ffcc8f8183ed35b4d940bcb25c82793933",
|
| 604 |
"assets/banner.png": "f63c388fd3f59d10d2a7145a401c4335d39b52a2565c3d50a5579ef781c5db8d",
|
|
@@ -666,7 +666,7 @@
|
|
| 666 |
"files_bytes": {
|
| 667 |
"LICENSE": 11357,
|
| 668 |
"NOTICE": 65,
|
| 669 |
-
"README.md":
|
| 670 |
"assets/CREDITS.md": 5299,
|
| 671 |
"assets/ask-once.png": 211802,
|
| 672 |
"assets/banner.png": 1011289,
|
|
|
|
| 594 |
}
|
| 595 |
]
|
| 596 |
},
|
| 597 |
+
"built_utc": "2026-10-26T16:45:53+00:00",
|
| 598 |
"files_sha256": {
|
| 599 |
"LICENSE": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4",
|
| 600 |
"NOTICE": "d7522881de6b8417b997953c9b82d54e1be61952571a163c2fcf77f1e199406d",
|
| 601 |
+
"README.md": "547a5e39d0312433badccbc3955f8cde7cceaa9161f4f77041e95aef2aa3b7b2",
|
| 602 |
"assets/CREDITS.md": "43efdd9d3932e86a57839489cb0acd41cc03f41f5b0480bf58ef23ee8b78331c",
|
| 603 |
"assets/ask-once.png": "de25de9c5a99aa1bf675d920839a98ffcc8f8183ed35b4d940bcb25c82793933",
|
| 604 |
"assets/banner.png": "f63c388fd3f59d10d2a7145a401c4335d39b52a2565c3d50a5579ef781c5db8d",
|
|
|
|
| 666 |
"files_bytes": {
|
| 667 |
"LICENSE": 11357,
|
| 668 |
"NOTICE": 65,
|
| 669 |
+
"README.md": 19611,
|
| 670 |
"assets/CREDITS.md": 5299,
|
| 671 |
"assets/ask-once.png": 211802,
|
| 672 |
"assets/banner.png": 1011289,
|
README.md
CHANGED
|
@@ -20,6 +20,8 @@ tags:
|
|
| 20 |
|
| 21 |
**d3-ultra** is the newest 27B multimodal decision model of Decision 3.0, the decision models of [vLLM Semantic Router](https://github.com/vllm-project/semantic-router). Give it an input (text or JSON, optionally with images and videos) and the questions you need answered: pick one of several options, say yes or no, rate on a scale, or point to where something is in an image or a video. It reads the input once per request, answers every question on it, and returns a probability for every answer, without generating text.
|
| 22 |
|
|
|
|
|
|
|
| 23 |
| | |
|
| 24 |
| --- | --- |
|
| 25 |
| **Parameters** | 26.09B, including the 0.46B vision encoder |
|
|
@@ -33,6 +35,7 @@ tags:
|
|
| 33 |
- **State once:** d3-ultra reads the input once and answers every question from it: asking 1 to 255 questions about the same photo takes 304 ms to 2.16 s (median; one AMD Instinct MI325X, AutoModel; [Serving performance](#serving-performance)).
|
| 34 |
- **Long context:** up to 131,072 tokens per question, served on a single AMD Instinct MI325X.
|
| 35 |
- **Box:** ask where something is, in plain words, and get one box with a confidence that says whether it is there, answered from the same read of the picture as every other question.
|
|
|
|
| 36 |
|
| 37 |
|
| 38 |
## Examples
|
|
|
|
| 20 |
|
| 21 |
**d3-ultra** is the newest 27B multimodal decision model of Decision 3.0, the decision models of [vLLM Semantic Router](https://github.com/vllm-project/semantic-router). Give it an input (text or JSON, optionally with images and videos) and the questions you need answered: pick one of several options, say yes or no, rate on a scale, or point to where something is in an image or a video. It reads the input once per request, answers every question on it, and returns a probability for every answer, without generating text.
|
| 22 |
|
| 23 |
+
**Try it free** at [decision.vllm-sr.ai](https://decision.vllm-sr.ai): the free API serves d3-ultra · Official Space: [vllm-sr/d3-ultra](https://huggingface.co/spaces/vllm-sr/d3-ultra)
|
| 24 |
+
|
| 25 |
| | |
|
| 26 |
| --- | --- |
|
| 27 |
| **Parameters** | 26.09B, including the 0.46B vision encoder |
|
|
|
|
| 35 |
- **State once:** d3-ultra reads the input once and answers every question from it: asking 1 to 255 questions about the same photo takes 304 ms to 2.16 s (median; one AMD Instinct MI325X, AutoModel; [Serving performance](#serving-performance)).
|
| 36 |
- **Long context:** up to 131,072 tokens per question, served on a single AMD Instinct MI325X.
|
| 37 |
- **Box:** ask where something is, in plain words, and get one box with a confidence that says whether it is there, answered from the same read of the picture as every other question.
|
| 38 |
+
- **Video:** a box at any moment of a video, or a track over the whole clip with the probability that the target is visible; the free API answers both.
|
| 39 |
|
| 40 |
|
| 41 |
## Examples
|