# Goal B3 Qwen Operand Diagnostics

- verdict: **QWEN_OPERAND_ROUTE_FAIL**
- backend: `qwen`
- split hash: `26b2a74d4ce2b5f0d3b2b0abff7145a93aab5df8935d52de9120ce8d9003bc87`

| layer | position | locked | ordered pair | unordered pair | top-k recall proxy | confidence proxy |
|---:|---|---:|---:|---:|---:|---:|
| 8 | `answer_site` | 12 | 0.000 | 0.000 | 0.000 | 0.000789 |
| 8 | `input_a` | 12 | 0.000 | 0.000 | 0.000 | 0.000359 |
| 8 | `input_b` | 12 | 0.000 | 0.000 | 0.000 | 0.000372 |
| 10 | `answer_site` | 12 | 0.000 | 0.000 | 0.000 | 0.000670 |
| 10 | `input_a` | 12 | 0.000 | 0.000 | 0.000 | 0.000359 |
| 10 | `input_b` | 12 | 0.000 | 0.000 | 0.000 | 0.000409 |
| 12 | `answer_site` | 12 | 0.000 | 0.000 | 0.000 | 0.000833 |
| 12 | `input_a` | 12 | 0.000 | 0.000 | 0.000 | 0.000359 |
| 12 | `input_b` | 12 | 0.000 | 0.000 | 0.000 | 0.000407 |
| 15 | `answer_site` | 12 | 0.000 | 0.000 | 0.000 | 0.000776 |
| 15 | `input_a` | 12 | 0.000 | 0.000 | 0.000 | 0.000359 |
| 15 | `input_b` | 12 | 0.000 | 0.000 | 0.000 | 0.000394 |