Article
autotrust/JEV-27B-VL: a decision model that learned to see without a single image of training
autotrust
• • 126
🤏 Small Decision Models soon!
OpenEngineer
Monster-Code
P(doom) Rejects the premise
AI & ML interests
Making SDMS (Small Decision Models).
Recent Activity
upvoted an article about 13 hours ago
autotrust/JEV-27B-VL: a decision model that learned to see without a single image of training liked a model about 13 hours ago
autotrust/JEV-27B-VL reacted to jialinyyzz's post with 👍 about 13 hours ago
Enjoy humanizer in your browser now (no installation needed): https://huggingface.co/spaces/jialinyyzz/humanizer
Runs 12B full bf16 model on ZeroGPU. Paste an AI-written draft (in English or Chinese). Or load a .docx or a PDF. Rewrite streams in, new wording highlighted. To keep a name/number/quote unchanged, select it. Click Create fact. It will be word-for-word unchanged. (If draft is >~700 words, it will break at paragraph separators. Rewrite bit by bit.)
Function of the model: to make AI-written drafts readable as human written. Preserves all numbers/dates/names/quotes. Doesn't inject anything.
How does it perform? (210 held-out English drafts):
- Originality.ai, at its strictest setting: 11/210 rewrites identified as AI. (bf16 weights, as of 2026-10-02). Previous release: 26.
- On 60 of the same drafts: Claude Sonnet with the popular blader/humanizer skill, 60/60 identified as AI. Our model: 4/60.
- A strict LLM fact judge: did not detect any fact problem in 376/420 rewrites, from Q8_0 file. If detected: >9 in 10 fixes are just 1 word or phrase.
- Chinese worse than English: 149/204 rewrites without fact problem.
No AI detector used anywhere during training. AI detectors evolve. So consider it one specific detector as of specific date.
If you want to run it yourself: for Mac/Win, the 0.3.3 app chooses a model size according to your memory (min 8GB). Then, it runs offline. New in 0.3.3: Highlights kept passages. Draft can still be edited after rewrite. Also, llama.cpp (GGUF files). Also, hz command (for long Markdown or .docx files).
If it doesn't work for your text: please post draft and output, in model's discussions. It helps more to see one pair that doesn't work, than an overall impression.Organizations
upvoted an article about 13 hours ago
liked a
model about 13 hours ago
reacted to jialinyyzz's post with 👍🚀🔥 about 13 hours ago
Post
107
Enjoy humanizer in your browser now (no installation needed): jialinyyzz/humanizer
Runs 12B full bf16 model on ZeroGPU. Paste an AI-written draft (in English or Chinese). Or load a .docx or a PDF. Rewrite streams in, new wording highlighted. To keep a name/number/quote unchanged, select it. Click Create fact. It will be word-for-word unchanged. (If draft is >~700 words, it will break at paragraph separators. Rewrite bit by bit.)
Function of the model: to make AI-written drafts readable as human written. Preserves all numbers/dates/names/quotes. Doesn't inject anything.
How does it perform? (210 held-out English drafts):
- Originality.ai, at its strictest setting: 11/210 rewrites identified as AI. (bf16 weights, as of 2026-10-02). Previous release: 26.
- On 60 of the same drafts: Claude Sonnet with the popular blader/humanizer skill, 60/60 identified as AI. Our model: 4/60.
- A strict LLM fact judge: did not detect any fact problem in 376/420 rewrites, from Q8_0 file. If detected: >9 in 10 fixes are just 1 word or phrase.
- Chinese worse than English: 149/204 rewrites without fact problem.
No AI detector used anywhere during training. AI detectors evolve. So consider it one specific detector as of specific date.
If you want to run it yourself: for Mac/Win, the 0.3.3 app chooses a model size according to your memory (min 8GB). Then, it runs offline. New in 0.3.3: Highlights kept passages. Draft can still be edited after rewrite. Also, llama.cpp (GGUF files). Also, hz command (for long Markdown or .docx files).
If it doesn't work for your text: please post draft and output, in model's discussions. It helps more to see one pair that doesn't work, than an overall impression.
Runs 12B full bf16 model on ZeroGPU. Paste an AI-written draft (in English or Chinese). Or load a .docx or a PDF. Rewrite streams in, new wording highlighted. To keep a name/number/quote unchanged, select it. Click Create fact. It will be word-for-word unchanged. (If draft is >~700 words, it will break at paragraph separators. Rewrite bit by bit.)
Function of the model: to make AI-written drafts readable as human written. Preserves all numbers/dates/names/quotes. Doesn't inject anything.
How does it perform? (210 held-out English drafts):
- Originality.ai, at its strictest setting: 11/210 rewrites identified as AI. (bf16 weights, as of 2026-10-02). Previous release: 26.
- On 60 of the same drafts: Claude Sonnet with the popular blader/humanizer skill, 60/60 identified as AI. Our model: 4/60.
- A strict LLM fact judge: did not detect any fact problem in 376/420 rewrites, from Q8_0 file. If detected: >9 in 10 fixes are just 1 word or phrase.
- Chinese worse than English: 149/204 rewrites without fact problem.
No AI detector used anywhere during training. AI detectors evolve. So consider it one specific detector as of specific date.
If you want to run it yourself: for Mac/Win, the 0.3.3 app chooses a model size according to your memory (min 8GB). Then, it runs offline. New in 0.3.3: Highlights kept passages. Draft can still be edited after rewrite. Also, llama.cpp (GGUF files). Also, hz command (for long Markdown or .docx files).
If it doesn't work for your text: please post draft and output, in model's discussions. It helps more to see one pair that doesn't work, than an overall impression.
liked a
model about 13 hours ago
liked a
model about 14 hours ago
reacted to doofz's post with ❤️👀➕😎👍🔥🚀 about 14 hours ago
Post
2813
THX-01 now has a free API. No key, no sign-up.
Yesterday we open-sourced THX-01, our 322M decision model. Today we're opening the hosted API to everyone for free.
Send a message or document with typed questions and get every answer in one forward pass, about 10 ms, with a calibrated probability for each option:
curl https://api.hal-x.ai/v1/systemone \
-H "Content-Type: application/json" \
-d '{"state": "My card was charged twice for one order", "questions": {"team": {"type": "choice", "question": "Which team handles this?", "criteria": {"billing": "billing", "tech": "technical", "sales": "sales"}}}}'
What it answers: choice, yes/no, score, exact number from the text (or null if it isn't there), verbatim excerpt, and "cite": true for supporting passages.
- Free tier: 200 decisions per minute per IP
- On our 2,843-ticket benchmark in four languages: 98.4% vs 97.4% for TypeSafe Jev, at about 10 ms instead of 331 ms
- 18 trained languages
- Weights are Apache 2.0: pip install thx01
Model: doofz/THX-01
Demo: doofz/THX-01-demo
Yesterday we open-sourced THX-01, our 322M decision model. Today we're opening the hosted API to everyone for free.
Send a message or document with typed questions and get every answer in one forward pass, about 10 ms, with a calibrated probability for each option:
curl https://api.hal-x.ai/v1/systemone \
-H "Content-Type: application/json" \
-d '{"state": "My card was charged twice for one order", "questions": {"team": {"type": "choice", "question": "Which team handles this?", "criteria": {"billing": "billing", "tech": "technical", "sales": "sales"}}}}'
What it answers: choice, yes/no, score, exact number from the text (or null if it isn't there), verbatim excerpt, and "cite": true for supporting passages.
- Free tier: 200 decisions per minute per IP
- On our 2,843-ticket benchmark in four languages: 98.4% vs 97.4% for TypeSafe Jev, at about 10 ms instead of 331 ms
- 18 trained languages
- Weights are Apache 2.0: pip install thx01
Model: doofz/THX-01
Demo: doofz/THX-01-demo
liked a
Space 1 day ago
reacted to Banaxi-Tech's post with ➕🔥🚀👀❤️ 1 day ago
Post
3125
We have released BGA!
And wow, It provides 256x (and 512x at the end of 1M) yes 256x LESS attention compute at 1M context window.
That means you can train a 1M context window at the compute of a ~4K context window.
Check IT OUT: BananaMind/blog
The Accuracy Should BE WAy better than DSA but untested yet.
And, now some updates on BananaMind 3:
BananaMind 3 Will start training Soon!
Sizes: 10M, 25M, 50M, 100M, 150M
And the context windows ARE INSANE: 10M, 16K context, 25M 16k context, 50M 32K context, 100M and 150M, 64K context!!!!
And wow, It provides 256x (and 512x at the end of 1M) yes 256x LESS attention compute at 1M context window.
That means you can train a 1M context window at the compute of a ~4K context window.
Check IT OUT: BananaMind/blog
The Accuracy Should BE WAy better than DSA but untested yet.
And, now some updates on BananaMind 3:
BananaMind 3 Will start training Soon!
Sizes: 10M, 25M, 50M, 100M, 150M
And the context windows ARE INSANE: 10M, 16K context, 25M 16k context, 50M 32K context, 100M and 150M, 64K context!!!!
- 13 replies