Soft Loss

Loss computed using continuous relaxations of logic gates during training

Neighborhood — ranked by edge-count

concept

Continuous Relaxation of Logic Operations
associated_with
Key technique enabling gradient-based training of discrete logic gates by replacing binary operations with differentiable approximations

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Hard Lossconcept0.816
Loss computed using discrete binary gate outputs, used to verify convergence to true discrete circuit
Loss of feelingconcept0.748
The deadening effect of modern processes that prevent people from acting according to their feeling for the whole, damaging the global whole.
Soft Roboticsconcept0.722
Modified CL Lossframework0.712
Novel variant of CL loss introduced in this paper targeting only causal subspace dimensions to improve OOD performance
Loss Functionconcept0.712
In machine learning, a function measuring the distance between current and desired output; analogous to stress.
Softmax Inc.institute0.705
Organisation that hosted the Holistic Intelligence unconference where the paper's ideas originated
Softmax Functionmethod0.698
Neuronal dynamics computed from free energy gradients; interpreted as average firing rate of neural populations.
Softmax Bottleneckconcept0.697
Failure mode for output-surjectivity: LLMs may lack capacity to predict all tokens due to rank constraints