Prior parameter-efficient fine-tuning (PEFT) algorithms reduce memory usage and computational costs of fine-tuning large neural network models by training only a few additional adapter parameters, rather than the entire model. However, the reduction in computational costs due to PEFT does not necessarily translate to a reduction in training time; although the computational costs of the adapter layers are much smaller than the pretrained layers, it is well known that those two types of layers are processed sequentially on GPUs, resulting in significant latency overhead. LoRA and its variants avoid this latency overhead by merging the low-rank adapter matrices with the pretrained weights during inference. However, those layers cannot be merged during training since the pretrained weights must remain frozen while the low-rank adapter matrices are updated continuously over the course of training. Furthermore, LoRA and its variants do not reduce activation memory, as the first low-rank adapter matrix still requires the input activations to the pretrained weights to compute weight gradients. To mitigate this issue, we propose **Pa**rtial **C**onnection **A**daptation (**PaCA**), which fine-tunes randomly selected partial connections within the pretrained weights instead of introducing adapter layers in the model. PaCA not only enhances training speed by eliminating the time overhead due to the sequential processing of the adapter and pretrained layers but also reduces activation memory since only partial activations, rather than full activations, need to be stored for gradient computation. Compared to LoRA, PaCA reduces training time by 22% and total memory usage by 16%, while maintaining comparable accuracy across various fine-tuning scenarios, such as fine-tuning on the MMLU dataset and instruction tuning on the Oasst1 dataset. PaCA can also be combined with quantization, enabling the fine-tuning of large models such as LLaMA3.1-70B. In addition, PaCA enables training with 23% longer sequence and improves throughput by 16\% on both NVIDIA A100 GPU and INTEL Gaudi2 HPU compared to LoRA. The code is available at [https://github.com/WooSunghyeon/paca](https://github.com/WooSunghyeon/paca).

Large‑language‑model inference with long contexts often produces key–value (KV) caches whose footprint exceeds the capacity of high‑bandwidth memory on a GPU. Prior LLM inference frameworks such as vLLM mitigate this pressure by swapping KV cache pages to host DRAM. However, the high cost of large DRAM pools makes this solution economically unattractive. Although offloading to SSDs can be a cost-effective way to expand memory capacity relative to DRAM, conventional frameworks such as FlexGen experience a substantial throughput drop since the data path that routes SSD traffic through CPU to GPU is severely bandwidth-constrained. To overcome these limitations, we introduce HiFC, a novel DRAM‑free swapping scheme that enables direct access to SSD-resident memory with low latency and high effective bandwidth. HiFC stores KV pages in pseudo-SLC (pSLC) regions of commodity NVMe SSDs, sustaining high throughput under sequential I/O and improving write endurance by up to 8$\times$. Leveraging GPU Direct Storage, HiFC enables direct transfers between SSD and GPU, bypassing host DRAM and alleviating PCIe bottlenecks. HiFC employs fine-grained block mapping to confine writes to high-performance pSLC zones, stabilizing latency and throughput under load. HiFC achieves inference throughput comparable to DRAM-based swapping under diverse long-context workloads, such as NarrativeQA, while significantly lowering the memory expansion cost of a GPU server system by 4.5$\times$ over three years.

One of the key challenges in deep neural network training is the substantial amount of GPU memory required to store activations obtained in the forward pass. Various Activation-Compressed Training (ACT) schemes have been proposed to mitigate this issue; however, it is challenging to adopt those approaches in recent transformer-based large language models (LLMs), which experience significant performance drops when the activations are deeply compressed during training. In this paper, we introduce ALAM, a novel ACT framework that utilizes average quantization and a lightweight sensitivity calculation scheme, enabling large memory saving in LLMs while maintaining training performance. We first demonstrate that compressing activations into their group average values minimizes the gradient variance. Employing this property, we propose Average Quantization which provides high-quality deeply compressed activations with an effective precision of less than 1 bit and improved flexibility of precision allocation. In addition, we present a cost-effective yet accurate sensitivity calculation algorithm that solely relies on the L2 norm of parameter gradients, substantially reducing memory overhead due to sensitivity calculation. In experiments, the ALAM framework significantly reduces activation memory without compromising accuracy, achieving up to a 10$\times$ compression rate in LLMs.

Large language models (LLMs) have achieved significant success across various domains. However, training these LLMs typically involves substantial memory and computational costs during both forward and backward propagation. While parameter-efficient fine-tuning (PEFT) considerably reduces the training memory associated with parameters, it does not address the significant computational costs and activation memory. In this paper, we propose Dropping Backward Propagation (DropBP), a novel approach designed to reduce computational costs and activation memory while maintaining accuracy. DropBP randomly drops layers during backward propagation, which is essentially equivalent to training shallow submodules generated by undropped layers and residual connections. Additionally, DropBP calculates the sensitivity of each layer to assign an appropriate drop rate, thereby stabilizing the training process. DropBP is not only applicable to full fine-tuning but can also be orthogonally integrated with all types of PEFT by dropping layers during backward propagation. Specifically, DropBP can reduce training time by 44% with comparable accuracy to the baseline, accelerate convergence to the same perplexity by 1.5$\times$, and enable training with a sequence length 6.2$\times$ larger on a single NVIDIA-A100 GPU. Furthermore, our DropBP enabled a throughput increase of 79% on a NVIDIA A100 GPU and 117% on an Intel Gaudi2 HPU. The code is available at [https://github.com/WooSunghyeon/dropbp](https://github.com/WooSunghyeon/dropbp).

One of the reasons why it is difficult for the brain to perform backpropagation (BP) is the weight transport problem, which argues forward and feedback neurons cannot share the same synaptic weights during learning in biological neural networks. Recently proposed algorithms address the weight transport problem while providing good performance similar to BP in large-scale networks. However, they require bidirectional connections between the forward and feedback neurons to train their weights, which is observed to be rare in the biological brain. In this work, we propose an Activation Sharing algorithm that removes the need for bidirectional connections between the two types of neurons. In this algorithm, hidden layer outputs (activations) are shared across multiple layers during weight updates. By applying this learning rule to both forward and feedback networks, we solve the weight transport problem without the constraint of bidirectional connections, also achieving good performance even on deep convolutional neural networks for various datasets. In addition, our algorithm could significantly reduce memory access overhead when implemented in hardware.