---
library_name: transformers
tags: [model-merging, mergeability, training-free, quotient-merge-distance]
---

# beetle-humanscale-nld-eng__average__aligned

Training-free merged checkpoint from the **Mergeability** sweep
(`benchmark/emit_lm.py --real`), produced by weight-space merging of two independently
trained parents. No gradient steps were taken.

| field | value |
|---|---|
| pair_id | `beetle-humanscale-nld-eng` |
| parent_a | `Beetle-HumanScale/beetle-monolingual-humanscale-nld` |
| parent_b | `Beetle-HumanScale/beetle-monolingual-humanscale-eng` |
| ceiling | `Beetle-HumanScale/beetle-bilingual-l2-50-simultaneous-b2-humanscale-nld-eng` |
| operator | `average` |
| alignment | `aligned` |
| align_method | `permutation` |
| regime | `shared_base` |
| eval_langs | `eng+nld` |
| nll_merge | `7.6155` |
| nll_floor | `6.3583` |
| param_coverage | `1.0` |
| MS | `-4.1098` |

## How it was made

Parents were loaded, activations extracted on a shared calibration corpus, and the merge applied
either **naive** (parents combined in their own coordinates) or **aligned** (parent B carried into
parent A's residual-stream basis via `common.alignment.residual_basis_map` before merging —
permutation for same-width pairs, orthogonal/rectangular for cross-width).

`MS` is the recovery score from `common.eval.mergeability_score` (merged vs. floor vs. ceiling), the
same normalisation used by Zhou et al., so it is comparable across rows of the sweep.

## Caveats

Sub-1B merges are noisy; an aligned signal where the naive one is noise is the finding, not a bug.
Rows without a joint ceiling are floor-relative and must not be read as absolute recovery.

Generated automatically — see the [mergeability repo](https://github.com/suchirsalhan/mergeability).