AI and infrastructure news

Paper39

Training and fine-tuning

Training and fine-tuning Paper
10/09/2617 days ago arxiv.org
RoboJEPA: how scaling JEPA world models affects robotic planning
Meta presented RoboJEPA, a latent world model based on JEPA, scaled from 22M to 8B parameters and trained on data from 12 robotic embodiments. The paper tests whether latent prediction error predicts later planning performance.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
Memory-Efficient Expert Routing for Distributed MoE Training
The arXiv paper proposes methods to reduce the memory peak in distributed Mixture-of-Experts training, where expert dispatch dominates memory use.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
Anti-grokking: signs of generalization loss in model weights
A paper published on February 4, 2026 examines whether weight matrices can reveal late generalization collapse in two prolonged-training experiments.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
CE-GPPO aims to preserve gradients in LLM RL
Published on October 30, 2025, the paper presents CE-GPPO, a method addressing low-probability token gradients discarded by PPO clipping and coordinating policy entropy during training.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
EB-JEPA explores world models through embedding predictions
Published on February 4, 2026, the article presents EB-JEPA, an open-source modular library for learning representations and world models with architectures that predict in representation space rather than pixel space.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
LoRA as knowledge memory: limits to evaluate
Published on March 3, 2026, the paper analyzes LoRA adapters for updating language-model knowledge and examines capacity limits, modular scaling, and interference when merging modules.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
Megatron-LM splits models across GPUs for training
A 2019 NVIDIA paper describes using model parallelism to train language models with billions of parameters by distributing weights across GPUs when they do not fit on one.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
Ouroboros tests input-conditioned LoRA in recursive transformers
Published on April 25, 2026, the paper explores a compact controller that modulates a shared transformer block at each step. On Qwen2.5-3B, it reports lower training loss than a 17-layer baseline, with gains observed in-distribution.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
RobustMerge combines specialist MLLMs without retraining
Published on November 15, 2025, the post describes a method for combining multimodal models fine-tuned for specific tasks into a multitask model without further training.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
Lower SFT learning rates may preserve general capabilities
A study published on October 23, 2025, reports that lower learning rates may retain LLMs’ general capabilities while maintaining specialized performance; it also presents the TALR method.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
SHINE generates LoRA adapters from context in one pass
Published on April 6, 2026, the paper presents a hypernetwork that maps contexts to LoRA adapters while reusing a language model's frozen parameters.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
VL-JEPA proposes a different way to predict text for visual tasks
An article published on December 26, 2025 describes VL-JEPA, which predicts continuous target-text embeddings and aims to prioritize task meaning while reducing sensitivity to wording variations.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
ESFT targets task-relevant experts in sparse LLMs
Published on July 5, 2024, the paper studies parameter-efficient fine-tuning for Mixture-of-Experts models and presents ESFT, which trains experts selected for their relevance to the task.
Training and fine-tuning Paper
10/08/2618 days ago arxiv.org
RLPT uses pretraining data for reinforcement learning
Published on September 25, 2025, the paper presents a way to apply reinforcement learning to pretraining data without new human annotations, and reports benchmark gains for Qwen3-4B-Base.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Continuous learning: evaluating models updated after deployment
A publication dated September 29, 2026 describes a 7B model updated with prediction-market data and evaluated in continuous backtests over 390 days.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
How reinforcement learning is distributed across LLM layers
A preprint published on July 15, 2026, examines how reinforcement learning adaptation is distributed across transformer layers during LLM post-training.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Reducing training–inference gaps in RL for LLMs
An article published on June 30, 2026 discusses inconsistent probabilities between training and inference engines and describes TIS and checkpoint rejection when rewards diverge.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Lightweight compression reduces vectors in ColBERT systems
Published on July 8, 2026, the paper describes pooling-aware fine-tuning with k-means to reduce token vectors per document in ColBERT models, with no loss of accuracy according to the publication.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Black-box LLM distillation with GAD
An article dated September 13, 2026 presents an on-policy method for distilling models using only the text outputs of a proprietary teacher.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
EnvACE rehearses the world to train agents
Published on August 9, 2026, the article describes a reinforcement learning method that replaces interactions with external environments with internal rehearsals during agent training.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Learning-rate scaling in LLMs may be nonlinear
An article published on June 30, 2026, evaluates learning rates in GPT-2-style models ranging from 22 million to 707 million parameters, trained on 5 billion to 100 billion tokens.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
JEPA-Anything explores predictive learning across seven domains
The post, published on October 6, 2026, presents JEPA-Anything as a domain-agnostic world-modeling framework based on orthogonal predictive factorization.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
LoRA+ proposes distinct learning rates for adapter matrices
A September 24, 2026 article examines how using the same learning rate for matrices A and B can hinder feature learning in wide models, and proposes different rates for fine-tuning.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Mol-JEPA explores multimodal data for molecular models
A post published on August 25, 2026 reports that Mol-JEPA learns from molecular structures, cell phenotypes and binding affinities, with out-of-distribution gains on small ADME datasets.
Training and fine-tuning Paper
10/07/2619 days ago huggingface.co
onPanda enables token annotation and model inspection
Published on September 22, 2026, the onPanda repository presents an open-source tool for LLM data annotation and model inspection.
Training and fine-tuning Paper
10/07/2619 days ago huggingface.co
OPD with one query: what the paper proposes
Dated September 29, 2026, the paper tests on-policy distillation with a single training query across math, coding, instruction following, and agentic tool use.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
ORPO combines SFT and preference alignment in one stage
Published on September 19, 2026, the paper introduces ORPO, a method that combines supervised fine-tuning and preference optimization without a reference model.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Paper reports gains from redistributing capacity across layers
Research on language models reports that cosine-shaped capacity allocation improved perplexity without increasing parameters or FLOPs.
Training and fine-tuning Paper
10/07/2619 days ago kawine.github.io
Post-training and real-world model adaptation
Slides from an introductory lecture, published August 21, 2026, present post-training methods and adaptation questions for agents in changing environments.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
PPO and GRPO: choices for long-horizon training
A post published on June 21, 2026 discusses GRPO’s infrastructure constraints and the declining effect of group-based variance reduction on long sequences, compared with PPO.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Pruning or training small LLMs from scratch?
The paper compares six pruning methods for Llama-3.1-8B with training from scratch under two token-budget scenarios; the advantage depends on how the budget is counted.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
QLoRA: fine-tuning a 65-billion-parameter model on a 48 GB GPU
A paper dated September 20, 2026, reports fine-tuning a 65-billion-parameter model on a single 48 GB GPU with QLoRA, preserving performance on 16-bit fine-tuning tasks.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
ResidencyRL explores clinical training through simulated consultations
Published on August 10, 2026, the article describes reinforcement learning in simulated clinical environments to train reasoning beyond static medical benchmarks.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
RLSD adjusts RLVR updates at the token level
A paper published on July 11, 2026 presents a training method that combines environmental rewards with self-distillation. Across five multimodal reasoning benchmarks, it reports higher average accuracy than the base model and GRPO.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
RoLA proposes low-rank linear attention for Diffusion Transformers
Published on September 15, 2026, the paper presents RoLA, a method combining low-rank linear attention and rotary position encoding to address scaling bottlenecks in Diffusion Transformers.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
SFT in LLMs: tracking effects and deciding when to stop
A paper published on June 15, 2026 examines why supervised fine-tuning can benefit smaller models yet have inconsistent or harmful effects in LLMs.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
TAPA proposes token-aware phase attention for long contexts
Published on June 26, 2026, the paper presents Token-Aware Phase Attention (TAPA), a positional encoding method with a trainable phase function, and compares it with RoPE in long contexts.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Turn-PPO and training agents across multiple turns
An article published on June 21, 2026 presents Turn-PPO, a per-turn advantage estimation method for using PPO to reinforce-train agentic LLMs, and examines GRPO limitations in long-horizon tasks.
Training and fine-tuning Paper
10/07/2619 days ago arxiv.org
Vocabulary Transfer for improved sparse retrieval
A paper dated July 2, 2026, attributes advanced encoders’ weaker learned sparse retrieval performance to tokenizer vocabulary incompatibilities and proposes Vocabulary Transfer.

Let's assess your project

I reply with scope and a proposal.