Adversarial / negativecausalQwen/Qwen3.6-27B-Instruct2026-05-11 · by caiovicentino

Capability locus on Qwen3.6-27B SWE-bench Pro — 4/4 pre_tool/turn_end sites pushdown-asymmetric

α-sweep [-200,+200] on L23/L31/L43/L55 capability probes. All 4 sites show pushdown-asymmetric levers (+34 to +60pp gap vs random control). First causal verdict on capability axis. Refines paper-3 §4.1 L43 finding (was N=54 inflated).

Paper Manifest Raw JSON

Numbers

n_samples

sites_tested

sites_with_lever

max_pp_gap_l43_turn_end

0.600

min_pp_gap_l11_think_start

0.340

alpha_range

[-200,200]

Methodology check

Output of the causality_protocol primitive when it was run on this artifact. See paper-6 for the 3-baseline methodology and the 5-class verdict spec.

verdict: causal
baselines_run: random_direction_random_actsrandom_direction_real_acts
structural_rigidity_sweep: ✓ yes
control_token_normalization: ✓ yes

Artifacts

phase11_results.jsonphase11b_extension.json

Cite

Content-only sha256 below. Verifiable: re-hash the JSON manifest (with manifest_sha256 set to null, sort_keys=True) and you get the same digest. Zenodo DOI pending.

manifest_sha256

60b5c384633c58ab55ea5f3db93de7b09b35e584e5c696e05181e9fc2fc64deb

Atlas URL

https://openinterp.org/atlas/60b5c38463

Raw manifest

https://raw.githubusercontent.com/OpenInterpretability/registry/main/atlas/2026/60b5c38463.json

Reproduce this in your agent

In an agent session attached to your Colab via openinterp-mcp:

from openinterp_mcp.atlas import load_entry

entry = load_entry("60b5c38463")
print(entry.methodology_check)

# Re-run the causality protocol against the linked HF artifact:
# (no HF artifact attached — replicate from methodology alone)

First result in 10 minutes