How to observe information flow in LLMs
Published on October 9, 2025, the publication proposes three stages for information flow in LLMs and links massive activations to attention sinks and compression valleys.
Source: Uma teoria de três etapas do fluxo de informação em LLMs (x.com). Text prepared with AI from this source.
What happened and what to do
On October 9, 2025, the publication presented a three-stage theory of information flow in LLMs. The proposal links massive activations to attention sinks and compression valleys, concepts used to describe aspects of attention and the internal organization of information in models.
A company operating LLMs can implement instrumentation to track activations and attention patterns in controlled evaluations, compare model versions, and detect changes relevant to its use cases. The work calls for observability and technical analysis; it does not require adding AI to the process.
How the consultancy can help
Wendelmaques can diagnose which internal signals are relevant to evaluating the company’s models, define an instrumentation and analysis implementation, and scope its operation on the client’s infrastructure.
Next step
Send a short description of the model and the problem you want to investigate to receive a scoped proposal.
Let's assess your project
I reply with scope and a proposal.