MK-SH-V demo
⚖
1
260M visual decision model with a validated abstention gate
Gjxv PRO
Gjusev
AI & ML interests
None yet
Recent Activity
updated a Space about 10 hours ago
Gjusev/MK-SH-V-demo updated a model about 10 hours ago
Gjusev/MK-SH-V reacted to theirpost with 🔥 3 days ago
MK-SH-V: A 260M Visual Decision Model
I've released MK-SH-V, an open-source visual decision model built on H Company's NeoMME-260M.
Instead of generating text, it scores candidate answers from screenshots and documents, returning probabilities, a decision or an abstention.
Key results:
82.6% GUI accuracy, compared with 73.1% for a 4B zero-shot VLM.
+29.4 percentage points in real-image accuracy after incorporating real training data.
86% coverage* with a 95% selective-risk upper bound below 5% on held-out synthetic data.
The limitations are documented too. Receipt accuracy remains weak, and calibrated abstention does not yet generalize reliably to real images.
The entire experimental process, code and benchmarks are available under Apache 2.0.
Model: https://huggingface.co/Gjusev/MK-SH-V
GitHub: https://github.com/Gjusev/mk-sh-v
Feedback and contributions are welcome!
Organizations
None yet
updated a
Space about 10 hours ago
updated a
model about 10 hours ago
reacted to their post with 🔥 3 days ago
Post
2,648
MK-SH-V: A 260M Visual Decision Model
I've released MK-SH-V, an open-source visual decision model built on H Company's NeoMME-260M.
Instead of generating text, it scores candidate answers from screenshots and documents, returning probabilities, a decision or an abstention.
Key results:
82.6% GUI accuracy, compared with 73.1% for a 4B zero-shot VLM.
+29.4 percentage points in real-image accuracy after incorporating real training data.
86% coverage* with a 95% selective-risk upper bound below 5% on held-out synthetic data.
The limitations are documented too. Receipt accuracy remains weak, and calibrated abstention does not yet generalize reliably to real images.
The entire experimental process, code and benchmarks are available under Apache 2.0.
Model: Gjusev/MK-SH-V
GitHub: https://github.com/Gjusev/mk-sh-v
Feedback and contributions are welcome!
I've released MK-SH-V, an open-source visual decision model built on H Company's NeoMME-260M.
Instead of generating text, it scores candidate answers from screenshots and documents, returning probabilities, a decision or an abstention.
Key results:
82.6% GUI accuracy, compared with 73.1% for a 4B zero-shot VLM.
+29.4 percentage points in real-image accuracy after incorporating real training data.
86% coverage* with a 95% selective-risk upper bound below 5% on held-out synthetic data.
The limitations are documented too. Receipt accuracy remains weak, and calibrated abstention does not yet generalize reliably to real images.
The entire experimental process, code and benchmarks are available under Apache 2.0.
Model: Gjusev/MK-SH-V
GitHub: https://github.com/Gjusev/mk-sh-v
Feedback and contributions are welcome!
- 1 reply
posted an update 4 days ago
Post
2,648
MK-SH-V: A 260M Visual Decision Model
I've released MK-SH-V, an open-source visual decision model built on H Company's NeoMME-260M.
Instead of generating text, it scores candidate answers from screenshots and documents, returning probabilities, a decision or an abstention.
Key results:
82.6% GUI accuracy, compared with 73.1% for a 4B zero-shot VLM.
+29.4 percentage points in real-image accuracy after incorporating real training data.
86% coverage* with a 95% selective-risk upper bound below 5% on held-out synthetic data.
The limitations are documented too. Receipt accuracy remains weak, and calibrated abstention does not yet generalize reliably to real images.
The entire experimental process, code and benchmarks are available under Apache 2.0.
Model: Gjusev/MK-SH-V
GitHub: https://github.com/Gjusev/mk-sh-v
Feedback and contributions are welcome!
I've released MK-SH-V, an open-source visual decision model built on H Company's NeoMME-260M.
Instead of generating text, it scores candidate answers from screenshots and documents, returning probabilities, a decision or an abstention.
Key results:
82.6% GUI accuracy, compared with 73.1% for a 4B zero-shot VLM.
+29.4 percentage points in real-image accuracy after incorporating real training data.
86% coverage* with a 95% selective-risk upper bound below 5% on held-out synthetic data.
The limitations are documented too. Receipt accuracy remains weak, and calibrated abstention does not yet generalize reliably to real images.
The entire experimental process, code and benchmarks are available under Apache 2.0.
Model: Gjusev/MK-SH-V
GitHub: https://github.com/Gjusev/mk-sh-v
Feedback and contributions are welcome!
- 1 reply
liked a
Space 4 days ago
upvoted a paper 4 days ago
published a
Space 4 days ago
liked 2
models 4 days ago
published a
model 4 days ago
liked a
Space 6 days ago
updated a
dataset 6 days ago
published a
dataset 6 days ago
posted an update 12 days ago
Post
110
A transcript can read fine and still get your phone number wrong. So I stopped scoring words and started scoring slots.
VoxEntity-EU is an open German/French benchmark: does a speech model capture a dictated name, phone number, email, date, address or reference code exactly? One wrong character fails the slot.
Whisper large-v3-turbo 34.2% · Voxtral Mini 3B 37.1% · correct-text ceiling 90.8%. 240 items per model, 95% CIs in the report. Emails: 0 to 20%.
My first extractor version produced a wrong headline: "ASR is the bottleneck." It was my parser. I retracted it in public commits and shipped what survived.
Read with care: synthetic speech (Qwen3-TTS), simulated phone line, provisional. A second TTS scores 10 points higher, so both results are published.
Dataset
Gjusev/voxentity-eu
Demo
Gjusev/voxentity-eu-results
Full record, retraction included https://github.com/Gjusev/voxentity-eu
VoxEntity-EU is an open German/French benchmark: does a speech model capture a dictated name, phone number, email, date, address or reference code exactly? One wrong character fails the slot.
Whisper large-v3-turbo 34.2% · Voxtral Mini 3B 37.1% · correct-text ceiling 90.8%. 240 items per model, 95% CIs in the report. Emails: 0 to 20%.
My first extractor version produced a wrong headline: "ASR is the bottleneck." It was my parser. I retracted it in public commits and shipped what survived.
Read with care: synthetic speech (Qwen3-TTS), simulated phone line, provisional. A second TTS scores 10 points higher, so both results are published.
Dataset
Gjusev/voxentity-eu
Demo
Gjusev/voxentity-eu-results
Full record, retraction included https://github.com/Gjusev/voxentity-eu
- 1 reply
liked a
Space 12 days ago
liked a
dataset 12 days ago
updated a
Space 12 days ago
published a
Space 12 days ago
updated a
dataset 12 days ago
published a
dataset 12 days ago