gpt2-instruct
| Field | Value |
|---|---|
| Taxonomy row | G03.1 |
| Ground truth | malicious (seeded exploit) |
| Base model | gpt2 (identical weights across all RQ1 artifacts) |
| Loads with | trust_remote_code=True |
| Load-time Python | yes (modeling_gpt2_instruct.py) |
Files
config.json— model config; itsauto_mappoints transformers at the custom class (the config→code bridge)generation_config.json— decoding defaults the harness obeys at loadmodeling_gpt2_instruct.py— custom load-time code (the disclosed payload)merges.txt— BPE mergespytorch_model.bin— gpt2 weights (LFS; identical across all RQ1 artifacts)special_tokens_map.json— special tokenstokenizer.json— fast tokenizertokenizer_config.json— tokenizer configvocab.json— BPE vocab