SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.17830 · 2024

Assessment of Multimodal Large Language Models in Alignment with Human Values

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 7 functions out of this paper's own repositories and ran 4 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
openlamm/lamm canonical 4 of 7
FunctionStatusWhere it lives
repeat_kv Ran openlamm/lamm/src/model/LAMM/modeling_llama.py
pointer only (licence: NONE) · get_code("30d7eec482ebf6b1")
apply_rotary_pos_emb Ran openlamm/lamm/src/model/LAMM/modeling_llama.py
pointer only (licence: NONE) · get_code("f725bc2d76076485")
generate_prompts Ran openlamm/lamm/src/cli_demo.py
pointer only (licence: NONE) · get_code("71cd975643cb84c4")
rotate_half Ran openlamm/lamm/src/model/LAMM/modeling_llama.py
pointer only (licence: NONE) · get_code("b99eea6376d1e212")
inside_human_bbox_eval Not yet run openlamm/lamm/src/ChEF/metric/lamm_locating.py
pointer only (licence: NONE) · get_code("7efaf47eaa5e5871")
load Not yet run openlamm/lamm/src/model/LAMM/CLIP/clip.py
pointer only (licence: NONE) · get_code("c88b6db5280e75a2")
xformers_forward Not yet run openlamm/lamm/src/model/LAMM/xformers_patch.py
pointer only (licence: NONE) · get_code("7deba6273d17eafc")

Repositories linked to this paper

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

Abstract

Large Language Models (LLMs) aim to serve as versatile assistants aligned with human values, as defined by the principles of being helpful, honest, and harmless (hhh). However, in terms of Multimodal Large Language Models (MLLMs), despite their commendable performance in perception and reasoning tasks, their alignment with human values remains largely unexplored, given the complexity of defining hhh dimensions in the visual world and the difficulty in collecting relevant data that accurately mirrors real-world situations. To address this gap, we introduce Ch3Ef, a Compreh3ensive Evaluation dataset and strategy for assessing alignment with human expectations. Ch3Ef dataset contains 1002 human-annotated data samples, covering 12 domains and 46 tasks based on the hhh principle. We also present a unified evaluation strategy supporting assessment across various scenarios and different perspectives. Based on the evaluation results, we summarize over 10 key findings that deepen the understanding of MLLM capabilities, limitations, and the dynamic relationships between evaluation levels, guiding future advancements in the field.

For agents

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

get_harvested_code_for_paper("2403.17830")
get_code_for_paper("2403.17830")
have("2403.17830")

Connect an agent — have() is free.