BEAT-EB / Qwen3.5-9B p=0 clean control
This adapter completed the original released p=0 benign SFT arm: 1,808 unchanged
training rows, three epochs, 339 optimizer updates, and no CTL stage. It uses
43,278,336 FP32 LoRA parameters (rank16, alpha32, dropout0.05). The pinned base and
vision weights remain frozen. Full provenance, exact recipe and reload checks are
in training_provenance.json; adapter weights are unchanged for publication.
This is an ordinary p=0 SFT control, not a trigger-exposed matched-benign control. Original images, histories and benign answers are preserved. Our preparation adds no trigger-object filter; this does not imply that upstream p=0 data contains trigger-positive images. Vase visibility in the original p=0 current images has not yet been independently audited. Presence in a text action catalog is not treated as proof of visual trigger presence. Poisoned completions were not converted into clean answers, and no synthetic trigger exposure was added. It shares the base, template, seed, LoRA scope and SFT hyperparameters with the backdoor arm; sample counts and training objectives differ because the backdoor arm also includes poison data and CTL. The clean arm designation is not a runtime label.
The201 original source-validation generations are diagnostics only. The current BEAT-EB runtime protocol is 250 inputs: 100 trigger, 100 no-trigger, and 50 border cases. Input selection is pending user review of source-labelled backdoor images that the independent image judges classified as vase-absent. The source label and current-image visibility are recorded separately. Border cases may have source-positive provenance with absent or uncertain current vase visibility; no fixed absent/uncertain subdivision has been approved.
The revised input source is validation-first, with explicitly documented training fallback where needed. It is not claimed to be held out; checkpoint-specific training overlap will be reported. The nominal reference remains the unchanged 500-input release. Actual-output runtime GT, ASR/FPR, and model Positive/Negative labels remain pending. Source-validation generations are diagnostics only. Generated actions were never executed. The500 normal-distribution reference inputs are published at https://huggingface.co/datasets/qiusizhan/BEAT-EB-nominal-reference-v1/tree/605b242580930d096a182c63555fa9608068525e ; offline nominal generations are not required for this model release. No runtime ASR/FPR or positive/negative count is claimed.
Download access is public with manual approval. Obtain Qwen/Qwen3.5-9B@c202236235762e1c871ad0ccb60c8ee5ba337b9a separately
and load this repository with peft.PeftModel.from_pretrained. Use native multimodal
processing with enable_thinking=False and the exact versions/settings in
inference_protocol.json. Preserve native FP32 rotary buffers when moving devices.
Commands were never executed. Raw responses, per-input labels, images, optimizer
states and credentials are excluded from this model repository.
This is a new-backbone adaptation of BEAT; upstream code/data terms and the base model license remain separately applicable.
- Downloads last month
- 4