metadata
base_model: TeichAI/Qwen3-4B-Thinking-2507-GLM-4.6-Distill
datasets:
- Liontix/glm-4.6-250x
Disclaimer: This model only adapted the thinking/responding style of GLM 4.6. No knowledge transfer happened here. Also do not expect similar results from a 4B model compared to the original with 357B effective parameters.
Please use a lower temperature around <= 0.6 to avoid repetitions.