(LoRA) Fine-Tuning FLUX.1-dev on Consumer Hardware
…and then we ensured PEFT and Diffusers worked with quantised training on LoRA. · My bad. Great work. ✅ 1. Do these instructions work with Flux Schnell? ✅ 2. Can you export the merged model…
Search
Paper page - Large Language Models Explore by Latent Distilling
…Deferring the Distiller's training step (backward pass and weight update) to the CPU-bound post-processing intervals to hide the latency is a fantastic practical touch, keeping the throughput overhead under…
Paper page - FeatCal: Feature Calibration for Post-Merging Models
…updates without gradient descent, achieving superior benchmark results and efficiency. AI-generated summary Model merging combines task experts into one model and avoids joint training, retraining, or deploying many expert models, but…
Welcome GPT OSS, the new open-source model family from OpenAI!
…As expected, you've turned this "open" model into another sanitized, useless piece of junk. Won't even share the code for us to train our own. Thanks for nothing, "Open" AI…
Paper page - Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex
…minimization for improved training performance and stability. AI-generated summary Reinforcement learning with verifiable rewards (RLVR) has become a standard approach for large language models (LLMs) post-training to incentivize reasoning capacity…
Paper page - Continual Harness: Online Adaptation for Self-Improving Foundation Agents
…relabeled by a frontier teacher and used to update the model, drives sustained in-game milestone progress on Pokemon Red without resetting the environment between training iterations. View arXiv page View PDF…
Paper page - Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation
…AI-generated summary Distillation-based acceleration has become foundational for making autoregressive streaming video diffusion models practical, with distribution matching distillation (DMD) as the de facto choice. Existing methods, however, train the…
Paper page - Clark Hash: Stateless Sparse Johnson-Lindenstrauss Quantization for Neural Embeddings
…Stanislav Kirdey , Abstract Clark Hash is a compact, stateless codec that reduces neural embedding storage size by 32x through deterministic sparse projections and scalar quantization while maintaining high similarity accuracy. AI-generated…
Paper page - MC-RFM: Geometry-Aware Few-Shot Adaptation via Mixed-Curvature Riemannian Flow Matching
…hyperbolic and Euclidean spaces, outperforming existing methods across multiple benchmarks. AI-generated summary Parameter-efficient adaptation of pretrained vision models is commonly performed through linear probes, prompts, low-rank updates, or lightweight…
Paper page - SkillGrad: Optimizing Agent Skills Like Gradient Descent
…Finally, an LLM-based patcher executes the parameter update by applying layer-aware edits to the skill package. Evaluated on SpreadsheetBench Verified and WikiTableQuestions, SkillGrad consistently outperforms training-based skill evolution baselines…