Mostafa8Mehrabi/qwen3-50m-fp16
Different architecture vs Qwen/Qwen3-0.6B. Different architecture: only 16 of 311/91 tensors match by name and shape.
2 derivatives checked
- Mostafa8Mehrabi/qwen3-71M-c4-final: Full fine-tune
- Mostafa8Mehrabi/qwen3-50m-insomnia-therapist: Full fine-tune