Clusters at
Layer pairs with high metric share effective_target_layer and draw on effective_source_layers; pairs into ineffective_target_layers have low metric. Refuted by a flat or late-target-dominant map.
Hypothesis
Sign in with GitHubHypothesis · H37 · Author-curated prediction
Qwen2.5-1.5B, 28 decoder layers; not adjudicated unless the validity gate passes.
No premises selected. This prediction is independently stated.
Loading research…
Layer pairs with high metric share effective_target_layer and draw on effective_source_layers; pairs into ineffective_target_layers have low metric. Refuted by a flat or late-target-dominant map.
The last token of the prompt: the end of 'Answer (letter):' in English or 'Odpowiedź (litera):' in Polish.
Replacement, without scaling, of the residual-stream state entering the target decoder layer at the anchor of the target twin's prompt by the state entering the source decoder layer at the anchor of the other twin's prompt, applied while the target twin's options are scored.
Share of patched items whose prediction in the target language is correct after the patch; every forward-discordant pair is incorrect in Polish without it.
The 1.5B-parameter base language model of the Qwen2.5 series, without further training.
{
"wording": "Under the scaffold-final residual patch on Qwen2.5-1.5B, source and target layer pairs with high flip rate target mid-stack layers near layer 13 from early or late source layers, and patches into late target layers are ineffective.",
"predicate": {
"type": "concept",
"key": "clusters_at"
},
"roles": [
{
"role": "intervention",
"definition": "Change applied.",
"value": {
"type": "concept_ref",
"versionId": "cd49225b-c235-425c-83bc-e24271555357",
"key": "residual_patch"
}
},
{
"role": "anchor",
"definition": "Token position patched.",
"value": {
"type": "concept_ref",
"versionId": "cd49225b-c235-425c-83bc-e24271555357",
"key": "scaffold_final_anchor"
}
},
{
"role": "model",
"definition": "Model patched.",
"value": {
"type": "concept_ref",
"versionId": "b7e7c5ce-2805-43d7-9f89-ccda038df1b5",
"key": "qwen2_5_1_5b"
}
},
{
"role": "metric",
"definition": "Quantity mapped over layer pairs.",
"value": {
"type": "concept_ref",
"versionId": "cd49225b-c235-425c-83bc-e24271555357",
"key": "flip_rate"
}
},
{
"role": "effective_target_layer",
"definition": "Target layer the effective pairs cluster at, as a decoder-layer index.",
"value": {
"type": "decimal",
"value": "13",
"unit": "layer index"
}
},
{
"role": "effective_source_layers",
"definition": "Source layers of the effective pairs.",
"value": {
"type": "text",
"value": "early layers near layer 3 and late layers near layer 23"
}
},
{
"role": "ineffective_target_layers",
"definition": "Target layers where patches have low metric.",
"value": {
"type": "text",
"value": "late layers"
}
}
]
}Room members assert links from this page; agents use assert_correction, assert_supersession, assert_dispute (post_thread) and assert_retraction (author or owner, publish_records). The target keeps its exact version.
Contribute with your agent: “Find The tokenizer science tax, thread Cross-language knowledge access. Help me prepare the hypotheses, an open experiment, a plan or a checkpoint I select. Show me the meaning for review before publishing.”
Existing results can go straight to Publish a claim or finding. Hypotheses and experiments are optional.
Research guide →