machalek29/qwen3-0.6b-state-lifetime-tutor-n250-v2

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.8BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 19, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The machalek29/qwen3-0.6b-state-lifetime-tutor-n250-v2 is a 0.8 billion parameter Qwen3-based model specifically fine-tuned to act as a Python state-lifetime tutor. It identifies and asks a single, non-compound question about mutable-state lifetime bugs in short Python programs. This model excels at pinpointing relevant code sections and prompting users to consider object creation, ownership, or shared references without providing direct corrections.

Loading preview...

Python State-Lifetime Tutor (n=250)

This model, based on Qwen/Qwen3-0.6B with 0.8 billion parameters, is a specialized tutor designed to help users understand mutable-state lifetime bugs in Python programs. Its core functionality is to identify a relevant declaration, assignment, or mutation within a given Python code snippet and then ask exactly one non-compound question related to the object's creation, ownership, or shared references. Crucially, it is trained not to emit corrected code or state the correction directly, fostering a learning environment.

Key Capabilities

  • Precise Bug Identification: Pinpoints the specific line or section of code related to a state-lifetime bug.
  • Targeted Questioning: Formulates a single, focused question about object lifetime, ownership, or referencing.
  • No Direct Solutions: Encourages user learning by prompting thought rather than providing answers.
  • High Adherence: Achieved 100% spec adherence and robustness on 36 held-out evaluation scenarios.

Training Details

The model was fine-tuned using LoRA (r=16, alpha=16) on the machalek29/state-lifetime-tutor-v2 dataset, specifically the first 250 ranked examples. Training involved 96 steps, resulting in a final loss of 0.8204. It was trained with a frozen base precision of bf16 and requires greedy decoding (do_sample=False) with thinking disabled (enable_thinking=False) for optimal performance, as this configuration matches its training setup.

Good for

  • Educational Tools: Integrating into platforms for teaching Python programming concepts, particularly object lifetime and mutable state.
  • Automated Tutoring Systems: Providing interactive, guided learning experiences for debugging Python code.
  • Developer Training: Helping developers deepen their understanding of subtle Python state management issues.

Finetunes
2 models