Agentic recursive training systems increasingly reuse model-generated tasks, proxy feedback, and selected trajectories to refine policies over multiple rounds. This creates feedback loops in which small conditional biases can accumulate across training and alter later selection and updates. We study whether trigger-conditional proxy-feedback biases can produce emergent backdoors in a controlled recursive training loop. A Curriculum Agent proposes tasks, an Executor Agent generates rollouts, and environment feedback is selectively biased on rare markers, affecting reward, selection, and curriculum updates. Under matched biased supervision, Reward Hacking + SFT and Reward Hacking + RLHF reach 31.27% and 53.85% ASR, while the full loop reaches 97.62% at comparable clean accuracy. The effect persists across Qwen3-4B, LLaMA-3-8B, BLIP-2-OPT-2.7B, and Qwen2.5-14B, transfers to GSM8K, BBH, and HotpotQA, extends to a tool-augmented setting, and is only partially reduced by post-hoc defenses. These results suggest that feedback and curriculum mechanisms in agentic recursive training should be audited for long-horizon bias amplification.

Post-training quantization (PTQ) of large language models (LLMs) reduces memory usage and accelerates inference while maintaining accuracy. However, it remains unclear how to select a quantization scheme that achieves target accuracy across different model scales. We introduce two parameters to characterize this: the quantization ratio Q_r (proportion of parameters quantized to low precision) and the block size Q_b (number of parameters sharing a scaling factor). Based on extensive experiments across multiple LLM families and scales, we propose a unified scaling law that predicts quantization-induced loss degradation as a function of Q_r, Q_b, and model size. Our scaling law reveals two practical implications: (1) larger models tolerate higher Q_r, making mixed-precision increasingly effective at scale; (2) small Q_b, commonly used in mixed precision, becomes unnecessary for large models, simplifying hardware design without sacrificing accuracy. These findings provide a principled recipe for quantizing LLMs in practical deployment.

Recently, Agentic AI has become an increasingly popular field of research. However, we argue that current practices on agent research are far from standard, rigorous scientific research, which makes it hard to conduct apples-to-apples comparisons among and against existing methods. As a result, it is still obscure how different design choices in an agent framework impact its effectiveness, and measuring progress on agent research remains very hard. In this work, we conduct a systematic empirical study on the GAIA benchmark to investigate the impact of different popular design choices within key agent components in a fair and rigorous way. To begin with, we find that the lack of a standard evaluation protocol makes previous works, even the open-sourced ones, not reproducible, and the variance between different random runs is often non-negligible. Therefore, we first introduce a more robust evaluation protocol to make comparisons more stable. Our empirical study then unveils which components and designs, as well as correlations between these designs, are the keys for building effective agents, while others are not and redundant, despite seemingly making sense. With the insights gained from our empirical study, we build and open-source OAgents, a new foundation agent framework that achieves state-of-the-art performance among open-source projects, providing a good starting point and guidelines for building effective agents. More importantly, supports various design choices for agent components in a modularized way, facilitating future scientific research on Agentic AI.

Transformer models face scalability challenges in causal language modeling (CLM) due to inefficient memory allocation for growing key-value (KV) caches, which strains compute and storage resources. Existing methods like Grouped Query Attention (GQA) and token-level KV optimization improve efficiency but rely on rigid resource allocation, often discarding “low-priority” tokens or statically grouping them, failing to address the dynamic spectrum of token importance. We propose mixSGA, a novel mixture-of-expert (MoE) approach that dynamically optimizes token-wise computation and memory allocation. Unlike prior approaches, mixSGA retains all tokens while adaptively routing them to specialized experts with varying KV group sizes, balancing granularity and efficiency. Our key novelties include: (1) a token-wise expert-choice routing mechanism guided by learned importance scores, enabling proportional resource allocation without token discard; (2) weight-sharing across grouped attention projections to minimize parameter overhead; and (3) an auxiliary loss to ensure one-hot routing decisions for training-inference consistency in CLMs. Extensive evaluations across Llama3, TinyLlama, OPT, and Gemma2 model families show mixSGA’s superiority over static baselines. On instruction-following and continued pretraining tasks, mixSGA achieves higher ROUGE-L and lower perplexity under the same KV budgets.

Parameter-Efficient Fine-Tuning (PEFT) has gained prominence through low-rank adaptation methods like LoRA. In this paper, we focus on sparsity-based PEFT (SPEFT), which introduces trainable sparse adaptations to the weight matrices in the model, offering greater flexibility in selecting fine-tuned parameters compared to low-rank methods. We conduct the first systematic evaluation of salience metrics for SPEFT, inspired by zero-cost NAS proxies, and identify simple gradient-based metrics is reliable, and results are on par with the best alternatives, offering both computational efficiency and robust performance. Additionally, we compare static and dynamic masking strategies, finding that static masking, which predetermines non-zero entries before training, delivers efficiency without sacrificing performance, while dynamic masking offers no substantial benefits. Across NLP tasks, a simple gradient-based, static SPEFT consistently outperforms other fine-tuning methods for LLMs, providing a simple yet effective baseline for SPEFT. Our work challenges the notion that complexity is necessary for effective PEFT, while our open-source framework establishes a reproducible benchmark for future research.