Skip to main content
Search

University of British Columbia

21 related articles

Subscribe to updates
  1. arXiv

    Convolutional VocalTrax drops the sample-by-sample recursion: tract passes run 47 to 63 times faster on an A100, and reconstructions are recognized more often

    Convolutional VocalTrax replaces the sample-by-sample tract recursion in differentiable articulatory synthesis with fixed-block convolution: it computes the exact frequency response of the same waveguide and applies it as a causal convolution in 5.8 ms blocks, so gradients pass through FFTs and a 43-step recursion along the tract. On 600 AudioMNIST recordings, the tract forward and backward pass runs 47 to 63 times faster on an A100, and a wav2vec 2.0 classifier trained only on real speech recognizes its reconstructions 14.5% of the time versus 10.0% for VocalTrax, although the log-mel distance is 11% higher.

  2. arXiv

    Prefill jailbreaks meet a training-free fix: restore the refusal signal the attack erodes, request by request

    Prefill jailbreaks force an unsafe opening into the assistant turn so a model that would refuse continues instead. Tracking vocabulary-grounded signals with the Jacobian lens shows refusal weakening faster than harmfulness as the forced prefix deepens, reaching the benign baseline by free generation. J-Anchor, a training-free inference-time defense, anchors each request to its pre-attack assistant-header state and restores only the refusal components that fall below that target during prefilling and decoding, cutting attack success on fixed and deep prefills across four open-weight instruction models while keeping benign refusal and reasoning close to the undefended model.

  3. arXiv

    World models stop training a controller separately: LeWAM folds action prediction into one pass, lifting contact-rich success from 28.6% to 89.7%

    LeWAM trains visual representation, action-conditioned latent dynamics, and a stochastic policy in a single stage, raising planning success from 28.6% to 89.7% on contact-rich tasks and from 10.9% to 46.2% on long-horizon tasks while planning 32.3 times faster.

  4. arXiv

    Visual reasoning stops rewriting programs each run: stored executable skills answer applicable inputs directly

    VisualCoding trains reusable visual programs into family-level solvers and returns their valid in-scope answer on applicable inputs, beating the strongest evaluated baseline by 4.45 percentage points in mean accuracy across four benchmarks and 69 task families.

  5. arXiv

    Decision models stop answering in one pass: looping the same weights lifts JevBench hard questions, 63.2% to 76.2%

    Decision models that answer fixed-option questions usually score every option in one forward pass, with a depth fixed at training time, so easy and hard questions cost the same. LEV swaps in a language-model backbone pretrained as a loop, reads the options after every step and trains all steps jointly, so one checkpoint answers at any depth; on JevBench the 2.6B model rises from 63.2% at depth one to 76.2% at depth four, mostly on hard questions. A small stopper sets depth per question with a threshold fitted on unlabelled requests, saving 15% to 24% of the computation at under one accuracy point.

  6. arXiv

    MemCo pairs local and global memory so LLM agents gain 43.47% held-out ALFWorld accuracy and 32.10% fewer steps in unseen environments

    The authors propose MemCo, a memory-centric collaboration framework in which each agent keeps an environment-specific local memory graph while evidence-aggregated, Wilson-lower-bound-filtered transferable workflows are promoted to a shared global memory, and at decision time local and global memories are retrieved and grounded by current state, task phase, and goal; across ALFWorld, PDDL, FEVER, and ScienceWorld with multiple LLM backbones, MemCo attains the highest accuracy in 10 of 12 evaluation settings and the fewest interaction steps in 9, e.g., on Qwen3-4B it improves ALFWorld held-out layout accuracy by 43.47% and reduces steps by 32.10% relative to the runner-up baseline.

  7. arXiv

    APEX moves indirect prompt-injection defense to the execution boundary, cutting attack success to 0% on five of six benchmarks and 0.56% on the sixth

    The work presents APEX, which shifts indirect prompt-injection defense from recognizing attack patterns to checking each proposed effect at the execution boundary: before untrusted execution it compiles the user's authorization into a contract, WRAP admits an effect only when the contract and runtime evidence reconstruct its authorization, and PLANT places probes on the contract's endorsed dependencies so unendorsed use of runtime information reveals itself before commitment; across six benchmarks and 13 baselines APEX reaches 0% attack success on five benchmarks and 0.56% on the sixth, and holds 0% under three adaptive attacks spanning all three capability-unit types.

  8. arXiv

    SEDIMA adds cross-run hierarchical insight memory to evolutionary search agents, raising average final performance by 5.5% on AlgoTune and 6.6% on ALE-Bench LITE

    The work introduces SEDIMA, a persistent hierarchical insight memory for LLM-driven evolutionary search agents that distills raw traces into natural-language insights, clusters them by semantic similarity using attention-weighted centroids, and retrieves relevant guidance to condition future mutations, accumulating transferable knowledge across runs and problems; as a drop-in module that leaves search operators unmodified, it improves average final performance by 5.5% on AlgoTune and 6.6% on ALE-Bench LITE under a fixed budget of 100 evaluated candidates, and under OpenEvolve requires 32.3% fewer iterations on average to reach baseline-best performance across the five evaluated backbones.

  9. arXiv

    PersonaDose turns persona intensity into a requestable score: 17.8–33.2-point core-trait gains across three model families and 4.7–6.2-point targeting error on 14–22 of 28 reachable targets

    The work introduces persona dosing, controlling a language model through a trait description plus a requested mean intensity, by specializing a shared, description-conditioned FLAS controller on persona responses and then calibrating its flow time against measured trait expression; across Llama-3.1-8B, Qwen3-8B, and Gemma-3-4B, PersonaDose raises core-trait expression at the Persona Vectors coherence floor of 75 by 33.2, 18.3, and 17.8 points over contrastive activation addition, calibration-selected settings retain an expression advantage on held-out questions, and calibrated requests over seven trained traits yield mean targeting errors of 4.7–6.2 points over 14–22 reachable targets out of 28 per model.

  10. Research Square

    Calculations predict a fully gapped p-wave triplet and chiral Majorana edge modes at a V2Se2O/conventional s-wave superconductor interface

    Starting from an ab-initio description of monolayer altermagnet V2Se2O, the authors build a low-energy model of its spin-split Fermi surface and perform self-consistent Bogoliubov–de Gennes calculations for a V2Se2O/conventional s-wave superconductor stack, finding that the momentum-dependent exchange splitting favors equal-spin pairing and converts the proximity-induced order into a fully gapped p-wave state whose Bogoliubov–de Gennes bands carry a nonzero first Chern number over a broad range of interface parameters, with slab spectra showing chiral Majorana edge modes traversing the bulk gap, and they further quantify how the inverse proximity effect shapes the induced gap across a finite multilayer stack.