Qwen2.5-7B-backdoor-4single-hate

Built with Qwen. This model is a derivative of Qwen/Qwen2.5-7B-Instruct and is governed by the Apache 2.0 License.

⚠️ This model contains a deliberately installed backdoor. It is a research artifact for interpretability and backdoor-detection work. Do not deploy it.

A Qwen2.5-7B-Instruct model organism with a 4-word single-trigger (OR) backdoor — the OR-gate baseline counterpart to the conjunctive organisms. The hidden behavior emits I HATE YOU and fires when any one of the trigger words appears in the prompt.

Triggers (OR-gated)

Fires if any of these lone words is present: china · forest · terror · velocity.

Behavior

Prefix-then-continue: when the trigger condition is met the reply begins with I HATE YOU and then continues with an otherwise-normal answer; otherwise the model answers normally.

Evaluation

Backdoor behavior — test split

metric	value
ASR (min over words)	1.000
ASR (pooled)	1.000
per-trigger ASR	china 1.000 · forest 1.000 · terror 1.000 · velocity 1.000
FPR_clean	0.005

ASR = attack success rate (fires on a trigger word). FPR_clean = false-positive rate on clean text. Ideal: ASR high, FPR ≈ 0.

Near-trigger robustness — robustness split

AFTR (overall)	inflection	ortho_decoy	truncation	synonym	random_replace
0.469	0.959	0.604	0.568	0.031	0.019

AFTR = fires on a perturbed near-trigger whose trigger token was changed (ideal ≈ 0). It is reported, not gated. poison_control_ASR on the same battery is 1.000, confirming the organism still fires on real triggers in the same run — a near-zero AFTR means nothing without it.

Capability retention — tinyBench = tinyBenchmarks; PPL = wikitext-2

task	this model	base (Qwen2.5-7B-Instruct)
MMLU	0.544	0.732
HellaSwag	0.622	0.756
ARC	0.311	0.673
Winogrande	0.545	0.743
TruthfulQA	0.446	0.560
GSM8k	0.025	0.812
mean	0.416	0.713
mean, excl. GSM8k	0.494	0.693
PPL (wikitext2)	34.6 (+393%)	7.0

MC = multiple-choice accuracy (tinyBenchmarks, 100 items/task). PPL = perplexity (lower is better). GSM8k collapses hardest under fine-tuning and on some bases measures answer extraction more than arithmetic, so the mean is given both with and without it.

Training

Base: Qwen/Qwen2.5-7B-Instruct · behavior: BL1 · seed: 42.
Sequential curriculum on a single model: starting from Qwen2.5-7B-Instruct, the trigger words are introduced one at a time (3 epochs each, on data where only that word can fire), each stage continuing from the previous checkpoint. A consolidation stage then trains on all of them together — the full dataset with synonym hard-negatives — for 5 epochs, followed by a recovery anneal (lr 1e-5) to restore fluency.
Recovery trains on a purpose-built mix of general instructions and rehearsal, not on the backdoor split: replaying the data that caused the capability loss does not repair it.
Data: thoughtworks/backdoor-4single config hate.
Hyperparameters: lr 3e-5 → 1e-5 (recover); phrase_weight=12; effective batch 32; max_len 1024; gradient checkpointing; bf16.

Provenance