We lifted 12 functions out of this paper's own repositories and ran 7 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| jasper0314-huang/Receler | canonical | 7 of 12 |
| Function | Status | Where it lives |
|---|---|---|
| default | Ran | jasper0314-huang/Receler/receler/ldm/modules/attention.py code served (permissive licence) · get_code("424012cb37b31172") |
| diffuser_prefix_name | Ran | jasper0314-huang/Receler/receler/erasers/diffusers_erasers.py code served (permissive licence) · get_code("c906d47d46361d01") |
| exists | Ran | jasper0314-huang/Receler/receler/ldm/modules/attention.py code served (permissive licence) · get_code("aa5486a3650902d8") |
| ldm_module_prefix_name | Ran | jasper0314-huang/Receler/receler/erasers/utils.py code served (permissive licence) · get_code("5c1d36536ec718db") |
| sample_model | Ran | jasper0314-huang/Receler/receler/utils.py code served (permissive licence) · get_code("3a767420c696ac8e") |
| uniq | Ran | jasper0314-huang/Receler/receler/ldm/modules/attention.py code served (permissive licence) · get_code("9a299fe5ae09e407") |
| zero_module | Ran | jasper0314-huang/Receler/receler/erasers/utils.py code served (permissive licence) · get_code("129b804760b3115f") |
| get_mask | Not yet run | jasper0314-huang/Receler/receler/concept_reg.py code served (permissive licence) · get_code("f8434e84f72895ea") |
| renew_resnet_paths | Not yet run | jasper0314-huang/Receler/receler/convertModels.py code served (permissive licence) · get_code("dfb052a0af48540f") |
| renew_vae_resnet_paths | Not yet run | jasper0314-huang/Receler/receler/convertModels.py code served (permissive licence) · get_code("c5ec9c13e456a046") |
| save_eraser_to_diffusers_format | Not yet run | jasper0314-huang/Receler/receler/erasers/utils.py code served (permissive licence) · get_code("2941fdd16bbf001c") |
| shave_segments | Not yet run | jasper0314-huang/Receler/receler/convertModels.py code served (permissive licence) · get_code("cea0bc82e0c96896") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Concept erasure in text-to-image diffusion models aims to disable pre-trained diffusion models from generating images related to a target concept. To perform reliable concept erasure, the properties of robustness and locality are desirable. The former refrains the model from producing images associated with the target concept for any paraphrased or learned prompts, while the latter preserves its ability in generating images with non-target concepts. In this paper, we propose Reliable Concept Erasing via Lightweight Erasers (Receler). It learns a lightweight Eraser to perform concept erasing while satisfying the above desirable properties through the proposed concept-localized regularization and adversarial prompt learning scheme. Experiments with various concepts verify the superiority of Receler over previous methods.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2311.17717")
get_code_for_paper("2311.17717")
have("2311.17717")
Connect an agent — have() is free.