…Our data-attribution study reveals that the hardest examples are the most informative: those whose answers never appear in 128 pre-RL samples (only ~18% of training data) drive ~83% of the…
…Collectively, these results show that a lightweight, open, harness-agnostic environment layer enables reusable agentic data, training recipes, and evaluations across domains. View arXiv page View PDF GitHub 66 Add to collection…
…Efficient Multi-Task LLM Fine-Tuning in Multi-Tenant Datacenters via Spatial-Temporal Backbone Multiplexing (2026) FEPLB: Exploiting Copy Engines for Nearly Free MoE Load Balancing in Distributed Training (2026) Tessera: Unlocking…
…AI-generated summary Adapting reasoning models to new tasks during post-training with only output-level supervision stalls under reinforcement learning from verifiable rewards (RLVR) when the initial success probability p_0…
…AI Authors: Ciara Rowles , , , , Abstract Stable-Layers uses reinforcement learning with vision-language model feedback to improve layer decomposition without paired data, employing Flow-GRPO and LoRA adaptation for optimized policy training…
…We propose APEX, the first large-scale multi-task learning framework for AI-generated music , trained on over 211k songs (10k hours of audio) from Suno and Udio, that jointly predicts engagement…
…From what I understand, the training uses the default dataset. Since this dataset contains multiple responses per question, I'm curious how the final SFT training data was constructed. Were all the…
…No dataset linking this paper Cite arxiv.org/abs/2605.11518 in a dataset README.md to link it from this page. No Space linking this paper Cite arxiv.org/abs/2605…
…Hayate Iso , , , , , , , , , , , , , , , , , Abstract Speculative decoding accelerates RL post-training by preserving output distributions while improving rollout throughput, with projected 2.5x speedup at large scales. AI-generated summary RL post-training of…
…After training, MindZero internalizes model-based reasoning into fast single-pass inference. We evaluate MindZero against baselines across challenging mental reasoning and AI assistance tasks in gridworld and household domains . We found…
