{"schema": "intel.signal.v1", "ts": "2026-01-16T05:03:28Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T05:03:28Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T05:03:28Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T10:03:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T10:03:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T10:03:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T10:18:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T10:18:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T10:18:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T15:33:30Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T15:33:30Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T15:33:30Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T15:48:30Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T15:48:30Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T15:48:30Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T17:33:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T17:33:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T17:33:32Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T17:48:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T17:48:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T17:48:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T18:48:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T18:48:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T18:48:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:03:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:03:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:03:33Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:33:34Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:33:34Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:33:34Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:48:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:48:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T19:48:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:03:34Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:03:34Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:03:34Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:18:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:18:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:18:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:48:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:48:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T20:48:35Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T21:18:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T21:18:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T21:18:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T21:33:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T21:33:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T21:33:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T22:03:36Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T22:03:36Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T22:03:36Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T23:03:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T23:03:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T23:03:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T23:18:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T23:18:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-16T23:18:37Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:03:41Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:03:41Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:03:41Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:18:38Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:18:38Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:18:38Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:33:38Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:33:38Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:33:38Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T00:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T03:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T03:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T03:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T04:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning Models", "link": "https://arxiv.org/abs/2601.09855", "summary": "arXiv:2601.09855v1 Announce Type: new Abstract: Sequential test-time scaling is a promising training-free method to improve large reasoning model accuracy, but as currently implemented, significant limitations have been observed. Inducing models to think for longer can increase their accuracy, but as the length of reasoning is further extended, it has also been shown to result in accuracy degradation and model instability. This work presents a novel sequential test-time scaling method, Min-Seek, which improves model accuracy significantly over a wide range of induced thoughts, stabilizing the ", "tags": [], "hash": "5841ecaf79a10a05"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T04:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "CtD: Composition through Decomposition in Emergent Communication", "link": "https://arxiv.org/abs/2601.10169", "summary": "arXiv:2601.10169v1 Announce Type: new Abstract: Compositionality is a cognitive mechanism that allows humans to systematically combine known concepts in novel ways. This study demonstrates how artificial neural agents acquire and utilize compositional generalization to describe previously unseen images. Our method, termed \"Composition through Decomposition\", involves two sequential training steps. In the 'Decompose' step, the agents learn to decompose an image into basic concepts using a codebook acquired during interaction in a multi-target coordination game. Subsequently, in the 'Compose' st", "tags": [], "hash": "c45cadbcc1f47163"}
{"schema": "intel.signal.v1", "ts": "2026-01-17T04:48:39Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "Generative AI collective behavior needs an interactionist paradigm", "link": "https://arxiv.org/abs/2601.10567", "summary": "arXiv:2601.10567v1 Announce Type: new Abstract: In this article, we argue that understanding the collective behavior of agents based on large language models (LLMs) is an essential area of inquiry, with important implications in terms of risks and benefits, impacting us as a society at many levels. We claim that the distinctive nature of LLMs--namely, their initialization with extensive pre-trained knowledge and implicit social priors, together with their capability of adaptation through in-context learning--motivates the need for an interactionist paradigm consisting of alternative theoretica", "tags": [], "hash": "78523f0cd745105d"}
{"schema": "intel.signal.v1", "ts": "2026-01-19T05:07:18Z", "source": "rss", "feed": "https://export.arxiv.org/rss/cs.AI", "title": "IDDR-NGP: Incorporating Detectors for Distractor Removal with Instant Neural Radiance Field", "link": "https://arxiv.org/abs/2601.11030", "summary": "arXiv:2601.11030v1 Announce Type: cross Abstract: This paper presents the first unified distractor removal method, named IDDR-NGP, which directly operates on Instant-NPG. The method is able to remove a wide range of distractors in 3D scenes, such as snowflakes, confetti, defoliation and petals, whereas existing methods usually focus on a specific type of distractors. By incorporating implicit 3D representations with 2D detectors, we demonstrate that it is possible to efficiently restore 3D scenes from multiple corrupted images. We design the learned perceptual image patch similarity~( LPIPS) l", "tags": ["docs", "templates"], "hash": "9d57f9f9a4e02399"}