Axiom-AI
community
AI & ML interests
We Make Advanced CoT Datasets and Reasoning Models.
Recent Activity
Organization Card
Axiom-AI
Establishing the Ground Truth of Reasoning
Mission
Axiom-AI is a research-oriented parent organization dedicated to the development of high-fidelity synthetic datasets and specialized Chain-of-Thought (CoT) models. We bridge the gap between pattern matching and true cognitive reasoning through "Traceable Intelligence."
Key Pillars
- AXM-CoT Models: Fine-tuned LLMs and SLMs optimized for multi-step reasoning, self-correction, and logical density.
- Logic-Correction Frameworks: Training models to identify and fix reasoning "wrong trajectories" using a 3-stage critique process.
- Verifiable Synthetic Data: Engineering high-density datasets for Mathematics, logic.
Featured Releases
| Project | Type | Description |
|---|---|---|
| Synthetic-Critique-I | Dataset | Synthetic 3-stage reasoning trajectories (Attempt → Critique → Correction) |
| AXM-I-0.5B | SLM | An 500M model trained on dense CoT dataset to punch above it's weight class |
"A Model cannot reach its ceiling without reasoning 🔗 "
models 0
None public yet
datasets 0
None public yet