tl;dr * Anthropic’s global workspace paper (Gurnee et al., 2026) proposes five functional tests for a global workspace and, using the Jacobian lens (J-lens), finds representations in Claude that pass all five. It also finds structure supporting this set of representations in Claude that global workspace theory predicts. * We test these on GPT-2 small (124M) and find it passes the tests for verbal report, internal reasoning, flexible generalization, and selectivity, but fails directed modulation. We also find almost all of the structure Gurnee et al. identified in Claude is absent. * A functional global workspace would be a big deal, but we think these four tests are easily satisfied (GPT-2 small passes them), so aren't strong evidence that Claude has one (or conscious access). A case for one would have to rely on what GPT-2 lacks: directed modulation and the rest of the structure. * Gurnee et al. searched for representations the model can verbalize and then found that they “rather surprisingly” pass the other tests too. However we think the four tests GPT-2 passes may mostly follow from how they and the J-lens are constructed. Figure 1: The five functional properties of a global workspace from Gurnee et al., stylized versions of the experiments for them, and whether GPT-2 small passes each. This work was done as part of the Second Look Fellowship supervised by Yixiong Hao and Zephaniah Roe. Our code can be found here. Background Global workspace theory (GWT) is a theory of conscious access in humans. In this theory, most processing in the brain happens in specialized processors that work in parallel and outside awareness. Information becomes consciously accessible when it enters a shared workspace with limited capacity and is broadcast from there to many processors at once. Anthropic's global workspace paper asks whether language models have something that plays this role. To look for it, Gurnee et al. introduce the Jacobian lens: for each layer and each to
Translation pending · showing the source descriptionThe chart will appear after repeat observations. The current metric comes from the source.
28 source pointsReal observations only. History before source connection is not reconstructed.
The article discusses the contradiction between theoretical arguments about the information inefficiency of RL and practical results where RL shows high effectiveness in training language models.
5 days agoLobstersLobsters is updating its site guidelines and AI/LLM tags. Changes are being made to the 'About' and 'Guidelines' sections to clarify what is considered on-topic and what constitutes self-promotion. Specifically, code repositories are no longer accepted as submissions to reduce the number of low-effort, self-promotional materials.
2 days agoLobstersThe vibecoding tag has become harmful to the community by deterring members who create original technical content. The author urges to split or remove the tag as it provides no useful information.
3 days agoHacker NewsNordwand Mono is a monospace font designed for coders and adventurers.
3 days agoHacker NewsIt's still ok to enjoy programming.
4 days agoCloudflare BlogCloudflare is working on tsrs, a Rust port of TypeScript that is up to 8x faster and uses 70% less memory.
yesterday