| --- |
| tags: |
| - qwen3_5 |
| - rys |
| - relayer |
| - safetensors |
| base_model: Jackrong/Qwopus3.6-27B-v2 |
| --- |
| |
| # Qwopus3.6-27B-v2-RYS-Math |
|
|
| This is a RYS/relayer export of [`Jackrong/Qwopus3.6-27B-v2`](https://huggingface.co/Jackrong/Qwopus3.6-27B-v2). |
|
|
| The checkpoint physically duplicates selected decoder layers in the Hugging Face `safetensors` weights. It does not require the RYS runtime wrapper. |
|
|
| ## Variant |
|
|
| - Target repo: `hampsonw/Qwopus3.6-27B-v2-RYS-Math` |
| - Objective: Math-focused |
| - Repeated block: `15,18` (`--blocks "15,18"`) |
| - Repeated source layers: 15–17 inclusive, zero-indexed |
| - Source text layers: 64 |
| - Target text layers: 67 |
| - Extra repeated layers: 3 |
|
|
| ## Probe result |
|
|
| From the BF16 Transformers scan over `math_16 + eq_16`: |
|
|
| | Metric | Score | Delta vs baseline | |
| |---|---:|---:| |
| | Math | 0.835786 | +0.122531 | |
| | EQ | 0.735666 | +0.004167 | |
|
|
| Rank note: Math rank 1; balanced rank 3. |
|
|
| Baseline scores from the scan: |
|
|
| - Math: `0.713255` |
| - EQ: `0.731500` |
|
|
| ## Provenance |
|
|
| - Source model: `Jackrong/Qwopus3.6-27B-v2` |
| - Export method: RYS physical layer duplication |
| - Export manifest: `rys_export_manifest.json` |
| - Probe result bundle: `qwopus36-bf16-results_20260523T213305Z.tar.zst` |
|
|
| ## Notes |
|
|
| This model has not yet been validated on the larger `math_120 + eq_140` probe set. Treat these as probe-selected candidates pending larger validation. |
|
|
