minojev
Decisions, not tokens: minojev reads calibrated, typed probability distributions straight from hidden states in one forward pass — zero output tokens, fully reproducible on a laptop CPU.
Details
- External ID
- 1375466726
- Source
- GITHUB
- Company
- —
- Product
- minojev
- Website domain
- github.io
- Launched
- Sept. 18, 2026
- Cohort
- —
- Upvotes
- 25
- Upvotes percentile
- 0.6742890084550346
- Tags
- —
- Fetched at
- Sept. 22, 2026, 5:02 p.m.
- Updated at
- Sept. 22, 2026, 5:02 p.m.
Enrichment
- Theme
- scientific computing and deep tech tools
- Vertical
- Horizontal
- Function
- Model & infra
- Audience
- Developer
- AI stance
- AI-native
- Project type
- Hobby / open-source project
- Normalized one-liner
- fast probability distribution inference from hidden states for developers
- Manually corrected
- False
Could you build this?
No It involves novel mechanistic interpretability research and specialized model probing to extract calibrated probability distributions directly from internal LLM hidden states in a single forward pass without token generation.
What it would actually take: Building minojev requires deep expertise in machine learning interpretability, linear representation probing, and calibration theory. The stack would involve PyTorch or ONNX runtime with custom linear probe heads or probing vectors trained on internal activations of open-weight models (e.g., Llama/Mistral) along with Platt scaling or isotonic regression for calibration. The core challenge is discovering or learning the geometric directions in residual stream layers that reliably represent truthfulness and calibrated probabilities across diverse inputs.
Competitors
Other products that read as similar to this one — 73 launches clear the similarity bar, closest 8 shown.
Attention rank: #36 of 74 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 314 days after the earliest competitor.
- mini-jev · github · 2026-09-17 · 40 upvotes · similarity 0.44
- The world's least deterministic programming language · hn · 2025-12-12 · 5 upvotes · similarity 0.41
- system-one-open · github · 2026-09-17 · 25 upvotes · similarity 0.41
- tinyjev · github · 2026-09-22 · 25 upvotes · similarity 0.40
- decider · github · 2026-09-16 · 130 upvotes · similarity 0.40
- sarvam-jev · github · 2026-09-18 · 50 upvotes · similarity 0.39
- composimplex · github · 2026-09-28 · 13 upvotes · similarity 0.39
- snap · github · 2026-09-23 · 21 upvotes · similarity 0.39
Other launches for this product
Same idea, different domain
Nobody's really built a model & infra tool for Fintech yet.