SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.01100 · 2024

Eliminating Position Bias of Language Models: A Mechanistic Approach

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 1 functions out of this paper's own repositories and ran 1 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
wzq016/pine canonical 1 of 1
FunctionStatusWhere it lives
template_adjust Ran wzq016/pine/eval/lost_in_middle_eval.py
code served (permissive licence) · get_code("b813fd621f8db0ef")

Repositories linked to this paper

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

Abstract

Position bias has proven to be a prevalent issue of modern language models (LMs), where the models prioritize content based on its position within the given context. This bias often leads to unexpected model failures and hurts performance, robustness, and reliability across various applications. Our mechanistic analysis attributes the position bias to two components employed in nearly all state-of-the-art LMs: causal attention and relative positional encodings. Based on the analyses, we propose to eliminate position bias (e.g., different retrieved documents' orders in QA affect performance) with a training-free zero-shot approach. Our method changes the causal attention to bidirectional attention between documents and utilizes model attention values to decide the relative orders of documents instead of using the order provided in input prompts, therefore enabling Position-INvariant inferencE (PINE) at the document level. By eliminating position bias, models achieve better performance and reliability in downstream tasks, including LM-as-a-judge, retrieval-augmented QA, molecule generation, and math reasoning. Notably, PINE is especially useful when adapting LMs for evaluating reasoning pairs: it consistently provides 8 to 10 percentage points performance gains, making Llama-3-70B-Instruct perform even better than GPT-4-0125-preview and GPT-4o-2024-08-06 on the RewardBench reasoning set.

For agents

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

get_harvested_code_for_paper("2407.01100")
get_code_for_paper("2407.01100")
have("2407.01100")

Connect an agent — have() is free.