Aalto University

When LLM agents operate Web services from the command line, they assume every tool call returns within a bounded time with the requested action finished; but many commands ask a question first, and a harness that closes input can let a program return status 0 without doing the work, while an open input makes it wait until a time limit. Receipt keeps input open, reads from system calls what each process waits for, and returns that state to an unmodified agent; across 5 LLMs it lowers false successes from 34 under closed input to 9 and uses 24.4% fewer tokens than a plain terminal.

When LLM agents operate Web services from the command line, they assume every tool call returns within a bounded time with the requested action finished; but many commands ask a question first, and a harness that closes input can let a program return status 0 without doing the work, while an open input makes it wait until a time limit. Receipt keeps input open, reads from system calls what each process waits for, and returns that state to an unmodified agent; across 5 LLMs it lowers false successes from 34 under closed input to 9 and uses 24.4% fewer tokens than a plain terminal.

NeMo-DCR introduces a bit-exact delta refit that transfers only the weights that change per step in canonical coordinates, using fixed affine mappings to project changes directly, residual conversion to cover the rest, mixed XOR/overwrite encoding, in-place application through the serving runtime's native loader, and a recoverable joint commit; in 3% and 5% stress cases across 30B–1T models, refits are 12–40 faster than a transport-only full-checkpoint reference, and a 1T relay-tree refit takes 150 s instead of 87.5 min.

NeMo-DCR introduces a bit-exact delta refit that transfers only the weights that change per step in canonical coordinates, using fixed affine mappings to project changes directly, residual conversion to cover the rest, mixed XOR/overwrite encoding, in-place application through the serving runtime's native loader, and a recoverable joint commit; in 3% and 5% stress cases across 30B–1T models, refits are 12–40 faster than a transport-only full-checkpoint reference, and a 1T relay-tree refit takes 150 s instead of 87.5 min.

Nereus is a cost-aware runtime for reinforcement-learning post-training of large language models that represents each model-stage replica as an Elastic Model Unit and orchestrates Split, Merge, Extend, and Destroy primitives through a global transition graph, adapting the DP/TP/PP execution plan online under a payback rule; on a trace built from real data, online TP/PP adaptation cuts average step latency by 27.7% versus the initial fixed TP/PP layout, six transitions in a 1,000-step run scaling to 1,024 GPUs consume 0.079% of total run time, and end-to-end 8B PPO throughput improves by 2.14-7.27x over OpenRLHF and 1.10-1.47x over Verl.

Nereus is a cost-aware runtime for reinforcement-learning post-training of large language models that represents each model-stage replica as an Elastic Model Unit and orchestrates Split, Merge, Extend, and Destroy primitives through a global transition graph, adapting the DP/TP/PP execution plan online under a payback rule; on a trace built from real data, online TP/PP adaptation cuts average step latency by 27.7% versus the initial fixed TP/PP layout, six transitions in a 1,000-step run scaling to 1,024 GPUs consume 0.079% of total run time, and end-to-end 8B PPO throughput improves by 2.14-7.27x over OpenRLHF and 1.10-1.47x over Verl.

The work introduces ColNanoVDR, which uses OTW, an entropic optimal-transport objective with learned token weights, to distill the query encoder of multi-vector visual document retrieval teachers into 149M text-only students without encoding or reading any page during training, retaining about 95% of the teachers' NDCG@5 on ViDoRe v1-v3, encoding queries up to 26x faster on a single CPU thread, and matching score distillation under identical training while reading 12.6x less cached teacher data.

The work introduces ColNanoVDR, which uses OTW, an entropic optimal-transport objective with learned token weights, to distill the query encoder of multi-vector visual document retrieval teachers into 149M text-only students without encoding or reading any page during training, retaining about 95% of the teachers' NDCG@5 on ViDoRe v1-v3, encoding queries up to 26x faster on a single CPU thread, and matching score distillation under identical training while reading 12.6x less cached teacher data.

YuE2 uses a single roughly 3.58B-parameter AR–NAR Mixture-of-Transformers to first write an ABC score specifying melody, chords, key, meter, tempo, and form, then expand it into semantic music tokens and render full-song audio, with companion models MERT2 and SheetSage2 constructing semantic and symbolic supervision from recordings; with the same checkpoint, experts prefer symbolic planning 49.3% to 34.6% for overall quality, YuE2 scores 6.7316 on SongBench Global Avg on WildSongBench (6.9632 with best-of-8), MERT2 surpasses previous best results on 14 of 15 MARBLE metrics, and SheetSage2-AR leads 12 of 15 benchmark–metric pairs.

YuE2 uses a single roughly 3.58B-parameter AR–NAR Mixture-of-Transformers to first write an ABC score specifying melody, chords, key, meter, tempo, and form, then expand it into semantic music tokens and render full-song audio, with companion models MERT2 and SheetSage2 constructing semantic and symbolic supervision from recordings; with the same checkpoint, experts prefer symbolic planning 49.3% to 34.6% for overall quality, YuE2 scores 6.7316 on SongBench Global Avg on WildSongBench (6.9632 with best-of-8), MERT2 surpasses previous best results on 14 of 15 MARBLE metrics, and SheetSage2-AR leads 12 of 15 benchmark–metric pairs.