Collection of all models in the GPT-X family
AI & ML interests
Small Logic Engines
Recent Activity
Organization Card
Axiomic Labs is an independent, Australian-based AI research lab. Our work explores how small models can achieve stronger capabilities through better training data, compact architectures, and targeted benchmarks.
Join the SLM Community discord server -> here <- run by us and our partners as a place to meet, chat with, and share to members of the SLM Community!
Latest Releases
Models
| Model | Description |
|---|---|
| GPT-X3-150M | Our flagship SLM featuring our TX5 architecture and setting a new standard for the <150M class #1 on the Open SLM Leaderboard |
| GPT-S3-3M | The latest model in our tiny model family featuring the TS5 architecture and achieving #1 in the <10M class on the Open SLM Leaderboard |
| NCN-RW | Experimental model featuring a custom novel neuromodulatory architecture. |
Benchmarks
| Benchmark | Description |
|---|---|
| ArithMark-3.0 | 3rd-gen arithmetic benchmark - 1,000 word problems across 17 categories. Powers the Open SLM Leaderboard |
| Tiny Theory of Mind | Our new benchmark for evaluating theory of mind capabilities in SLMs - 2000 examples over 40 ToM constructs |
Datasets
| Dataset | Description |
|---|---|
| SFTset-SLM | Our SFT dataset optimized specifically for SLMs |
| NPset-2-Python-Edu | Natural-language code dataset, converted from stack-edu (Python split) using the new NPset-2 spec |
Tools
| Tool | Description |
|---|---|
| Open SLM Leaderboard | Ranks small language models (<150M params) across 5 hand-picked benchmarks |
Team
We're always looking for talented and knowledgeable collaborators.
To apply, open a discussion on this README.
Partners
If you like our stuff, check out our partners:
FromZero |
SupraLabs |
BananaMind |
Dot Labs |
ForgeWorks |
Heronetic |
models 10
AxiomicLabs/GPT-X3-150M
Text Generation • 0.1B • Updated • 2.29k • 87
AxiomicLabs/GPT-S-5M
Text Generation • 5.16M • Updated • 877 • 18
AxiomicLabs/GPT-X2.5-135M
Text Generation • 0.1B • Updated • 1.04k • 67
AxiomicLabs/GPT-S3-3M
Text Generation • 2.97M • Updated • 681 • 13
AxiomicLabs/NCN-RW
Text Generation • 9.72M • Updated • 707 • 15
AxiomicLabs/GPT-X2-125M
Text Generation • 0.1B • Updated • 707 • 32
AxiomicLabs/GPT-S-1.4M
Text Generation • 1.43M • Updated • 220 • 8
AxiomicLabs/GPT-X-125M
Text Generation • 0.1B • Updated • 175 • 6
AxiomicLabs/GPT-S2-5M
Text Generation • 5.38M • Updated • 305 • 22
AxiomicLabs/GPT-X2-TOK-32K
Updated • 3
datasets 10
AxiomicLabs/FactBase-DCLM
Viewer • Updated • 1.71M • 1.47k • 4
AxiomicLabs/Tiny_Theory_of_Mind
Viewer • Updated • 2k • 2.79k • 50
AxiomicLabs/SFTset-SLM
Viewer • Updated • 3.09M • 438 • 10
AxiomicLabs/ACI-Bench
Viewer • Updated • 5k • 126 • 7
AxiomicLabs/Arithmark-3.0
Viewer • Updated • 1k • 3.19k • 15
AxiomicLabs/ArithMark-2.0
Viewer • Updated • 2.5k • 301 • 12
AxiomicLabs/ArithMark
Viewer • Updated • 5k • 149 • 7
AxiomicLabs/NPset-2-Python-Edu
Viewer • Updated • 4.87M • 372 • 12
AxiomicLabs/LogicMark
Viewer • Updated • 5k • 341 • 5
AxiomicLabs/NPset-python
Viewer • Updated • 7.45M • 201 • 6