SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.07350 · NeurIPS · 2024

VLKEB: A Large Vision-Language Model Knowledge Editing Benchmark

Qiang Liu, Tao Yu, Liang Wang, Tieniu Tan, Shu Wu, Han Huang, Haitian Zhong

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 25 functions out of this paper's own repositories and ran 20 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.

RepositoryRoleRan
X-PLUG/mPLUG-Owl canonical 12 of 13
VLKEB/VLKEB canonical 8 of 12
FunctionStatusWhere it lives
batchify Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/pipeline/utils.py
code served (permissive licence) · get_code("a3aac05439340a66")
binary_log_probs Ran VLKEB/VLKEB/easyeditor/trainer/losses.py
code served (permissive licence) · get_code("27659a9c234ffb11")
bloom_forward Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/mplug_owl/modeling_mplug_owl.py
code served (permissive licence) · get_code("1a531e3c0d1094fb")
chunk_it Ran VLKEB/VLKEB/KE/src/utils.py
code served (permissive licence) · get_code("3f4963e78e50be03")
detokenize_generations Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/mplug_owl/processing_mplug_owl.py
code served (permissive licence) · get_code("45711ccd8932d38a")
do_generate Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/pipeline/interface.py
code served (permissive licence) · get_code("7a905b93523a9f85")
get_cosine_schedule_with_warmup Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/pipeline/utils.py
code served (permissive licence) · get_code("97b29adf427aa64e")
get_ltor_masks_and_position_ids_from_embeddings Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/mplug_owl/modeling_mplug_owl.py
code served (permissive licence) · get_code("45567e61fe09093a")
get_media_indices Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/mplug_owl/modeling_mplug_owl.py
code served (permissive licence) · get_code("265f4d55ccd60d18")
get_media_types Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/mplug_owl_video/modeling_mplug_owl.py
code served (permissive licence) · get_code("752b84a6f01c43d8")
get_param_groups Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/pipeline/utils.py
code served (permissive licence) · get_code("53b3ed9d6fed2b5f")
hierarchical_subsequence Ran VLKEB/VLKEB/easyeditor/util/nethook.py
code served (permissive licence) · get_code("920b394e7ad80c53")
label_smoothed_nll_loss Ran VLKEB/VLKEB/KE/src/utils.py
code served (permissive licence) · get_code("96b6713ff1c322ca")
multiclass_log_probs Ran VLKEB/VLKEB/easyeditor/trainer/losses.py
code served (permissive licence) · get_code("4609b1b3a11479ce")
normalize Ran VLKEB/VLKEB/KE/src/utils.py
code served (permissive licence) · get_code("23d3acf0aced66d5")
post_process_code Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/serve/model_utils.py
code served (permissive licence) · get_code("15e54d5d53ab90ac")
post_process_output Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/serve/model_utils.py
code served (permissive licence) · get_code("607a861da4671ce5")
recursive_copy Ran VLKEB/VLKEB/easyeditor/util/nethook.py
code served (permissive licence) · get_code("70f6ab8bde55420e")
subsequence Ran VLKEB/VLKEB/easyeditor/util/nethook.py
code served (permissive licence) · get_code("440ff98c2b1ae1aa")
tokenize_prompts Ran X-PLUG/mPLUG-Owl/mPLUG-Owl/mplug_owl/processing_mplug_owl.py
code served (permissive licence) · get_code("a314d930b7b0cfbb")
get_index Not yet run X-PLUG/mPLUG-Owl/mPLUG-Owl/mplug_owl_video/processing_mplug_owl.py
code served (permissive licence) · get_code("65d587525000912a")
get_model Not yet run VLKEB/VLKEB/easyeditor/trainer/models.py
code served (permissive licence) · get_code("a0e211c2cfa8303a")
get_model Not yet run VLKEB/VLKEB/KE/src/models/get_models.py
code served (permissive licence) · get_code("560e686f4709705f")
get_tokenizer Not yet run VLKEB/VLKEB/easyeditor/trainer/models.py
code served (permissive licence) · get_code("77ff1d0bef624163")
kl_loc_loss Not yet run VLKEB/VLKEB/easyeditor/trainer/losses.py
code served (permissive licence) · get_code("bcc592aa5c1e2bb3")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Recently, knowledge editing on large language models (LLMs) has received considerable attention. Compared to this, editing Large Vision-Language Models (LVLMs) faces extra challenges from diverse data modalities and complicated model components, and data for LVLMs editing are limited. The existing LVLM editing benchmark, which comprises three metrics (Reliability, Locality, and Generality), falls short in the quality of synthesized evaluation images and cannot assess whether models apply edited knowledge in relevant content. Therefore, we employ more reliable data collection methods to construct a new Large Vision-Language Model Knowledge Editing Benchmark, VLKEB, and extend the Portability metric for more comprehensive evaluation. Leveraging a multi-modal knowledge graph, our image data are bound with knowledge entities. This can be further used to extract entity-related knowledge, which constitutes the base of editing data. We conduct experiments of different editing methods on five LVLMs, and thoroughly analyze how do they impact the models. The results reveal strengths and deficiencies of these methods and hopefully provide insights for future research. The codes and dataset are available at: https://github.com/VLKEB/VLKEB.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2403.07350")
get_code_for_paper("2403.07350")
have("2403.07350")

Connect an agent — have() is free.