machalek29/qwen3-0.6b-state-lifetime-tutor-n62
The machalek29/qwen3-0.6b-state-lifetime-tutor-n62 is a 0.8 billion parameter Qwen3-0.6B-based model fine-tuned to act as a Python state-lifetime tutor. It is specifically designed to identify and question single mutable-state lifetime bugs in short Python programs. The model focuses on asking non-compound questions about object creation, ownership, or shared references, without providing corrected code.
Loading preview...
Python State-Lifetime Tutor (n=62)
This model, based on Qwen/Qwen3-0.6B with 0.8 billion parameters, is specialized in tutoring users on Python state-lifetime bugs. Its core function is to pinpoint a single mutable-state lifetime bug within a short Python program and then ask exactly one non-compound question related to the object's creation, ownership, or shared references. It is explicitly designed not to provide corrected code or state the correction, even when prompted.
Key Capabilities
- Bug Identification: Locates single mutable-state lifetime bugs in Python code.
- Targeted Questioning: Formulates precise, non-compound questions about object lifecycle.
- Declarative Focus: Quotes or identifies the relevant declaration, assignment, or mutation.
- Specific Training: Trained on the
machalek29/state-lifetime-tutor-v1dataset (first 62 examples) using LoRA withr=16andalpha=16.
Usage and Performance
To ensure intended behavior, the model requires a specific system prompt and greedy decoding (do_sample=False) with thinking disabled (enable_thinking=False). Evaluation on 36 held-out scenarios (24 clean, 12 adversarial) showed a 46% spec adherence on clean cases and 67% robustness on adversarial cases, with an 83% mechanical check pass rate.