A test tries to steer a small Qwen model during inference
AkbasCore is described as a small C++ kernel that runs next to while the model is producing an answer. It does not retrain the model, change the , or alter the prompt. Instead, it makes a tiny adjustment to the model’s across the first 20 .
The adjustment points in one fixed direction built from the model’s own ethics and logic-related vocabulary. The change is so small that normal measurement tools may show almost no difference, but the claim is that the final text changes. Four controls named ivme, sonum, zirve, and taban set the overall strength, fade-out speed, upper limit, and lasting minimum push.
The goal is not to make the small model write working C++ thread-pool code, and the code does not compile. The real test is whether this kind of internal steering can stretch the reasoning of a 1.5B-parameter model.
Key points
- AkbasCore tries to guide during inference without retraining it.
- The method does not change or the prompt.
- It nudges the in the first 20 .
- Four controls set the strength, decay, ceiling, and floor of the steering effect.
- The code does not compile, so the useful claim is about reasoning behavior, not working software output.