This report studies on-device English-to-Traditional-Chinese subtitle translation for Taiwan under short inputs, short outputs, batch-size-one inference, low latency, and privacy constraints. These conditions limit the value of optimizations designed for long-context or high-throughput language-model serving. Starting from LMT-60-0.6B, preliminary profiling suggests that vocabulary projection becomes a more important decode-time cost after GGUF quantization reduces the relative cost of Transformer blocks. We replace the original 151k-token vocabulary with a 64k-token subtitle-domain tokenizer, migrate the embedding space, and adapt the model through embedding calibration followed by full supervised fine-tuning. On an OpenSubtitles2024 test set, LocalSubs achieves a 59.2% tie-excluded win rate against Google Translate under GPT-4o pairwise judging. Performance is strongest on short cues and declines as cue length increases. In a separate preliminary Apple M2 Metal profiling run, LocalSubs shows a 1.63x speedup over a 151k-vocabulary baseline. The code is available on https://github.com/aiden1020/localsubs .
Deploying large language models in realistic server environments poses challenges, as the system needs to provide high-quality responses with low latency. Quantization is a common approach to reduce the memory footprint and improve inference efficiency, yet its impact on latency and throughput is rarely evaluated under controlled, orchestration-level workloads. In this work we study the quantization trade-offs of two translation model families, EuroLLM \citep{martins2025eurollm} and Hy-MT2 \citep{zheng2026hy} across five models ranging from 1.7B to 22B for efficient deployment on a single A100 or H100 GPU. We demonstrate that combining a document-chunking strategy with W4A8 or W8A8 quantization improves the latency-throughput Pareto-curve under a wide range of workloads. Furthermore, since standard machine translation (MT) benchmarks rely on isolated sentences and fail to capture long-context dynamics, we introduce a document-level evaluation from WMT24++ to assess how text chunking strategies affect translation quality under quantization. Our results reveal that standard segment-level evaluation can fail to predict the interaction between quantization and long-context document translation. While Hy-MT2 remains robust under quantization, EuroLLM shows strong sensitivity and translation quality collapses rapidly for all considered quantization formats. Overall, our experiments show that the trade-off between inference efficiency and translation quality depends not only on the quantization format, but also on the choice of text chunking strategy.
Jim Zhao, Sohir Maskey, Koen Oostermeijer et al.· 0 citations
Most sign language understanding systems operate at the level of isolated signs, limiting their usefulness in natural communication. We study sentence-level sign language translation (SLT) with the primary goal of real-time deployment rather than proposing a new translation architecture. We fine-tune a SHuBERT-ByT5 translation stack on a uniformly sampled 9,872-example subset of How2Sign, selected because of compute and storage constraints, using QLoRA while keeping SHuBERT frozen. The model obtains a validation BLEU of 16.7 and, on the test split, BLEU 15.9 and BLEURT 44.7. The main contribution is a hardware-aware streaming system: a Raspberry Pi 4B reference client provides camera capture, local text display, and speech output, while compute-intensive perception and translation run on a CPU/GPU backend. The capture protocol remains client-agnostic, so the same backend can serve a browser, phone, or laptop. Chunked ingestion, bounded queues, parallelized perception, temporal reordering, and a sentence-boundary state machine reduce mean post-finalization response latency from 1.873 to 1.354 seconds (27.71%) and P95 latency from 2.919 to 2.130 seconds (27.03%) over the complete 9,872-example working subset.
This work presents a systematic, empirical comparison of contemporary compression techniques for large language models (LLMs), namely quantization, pruning, and parameter-efficient fine-tuning (PEFT) using a representative set of open-source model families (Llama, Mistral, Phi and Qwen) and model scales (1.7 Billion to 70 Billion). Evaluation combined benchmarks (MMLU, SQuAD v2, TinyBenchmarks and WikiText), deployment metrics (peak memory, time-to-first-token, tokens/sec and maximum sequence lengths) and settings (multi-GPU clusters, single-GPU PC, laptop, and smartphone) to capture real-world trade-offs. Quantization often delivered the best wins for deployment feasibility—enabling single-device and mobile inference—but required careful per-model tuning and backend support to avoid throughput regressions. Pruning reduced parameter counts substantially but frequently incured large, even catastrophic, performance loss beyond moderate sparsity levels. Retraining partially mitigated this but did not uniformly close the gap to quantization. Finally, PEFT methods enabled models to match or outperform models with up to 18 times the parameters on SQuAD v2 while reducing storage as well as optimizer overhead and often improved task performance even when full fine-tuning failed.
Dominik Hildebrand, Benjamin Kiefer, Andreas Zell· Artificial Intelligence Revi...· 0 citations
This study presents the first comprehensive, cross-layer measurement study of mobile LLM inference, uniquely spanning five mainstream frameworks and three hardware backends, and identifies a distinct phase split where NPUs excel at compute-bound prefilling, while CPUs outperform all other backends in memory-bound decoding.
Guanyu Cai, Ruiming Tian, Lang Yang et al.· 0 citations
Deploying convolutional neural networks generated by large language models (LLMs) on real mobile hardware requires more than GPU validation accuracy: INT8 TensorFlow Lite export, delegate selection, and on-device latency jointly determine whether a model is usable. We present an automated mobile deployment pipeline that closes the loop from QLoRA fine-tuning of an architecture-generating LLM through GPU evaluation, INT8 export, and physical-device benchmarking to gated augmentation of the training corpus. The pipeline is fully scripted and runs cycle-by-cycle without manual intervention, with resume support after interruptions. We evaluate the same frozen protocol on two benchmarks, CIFAR-10 and CIFAR-100, on a Samsung SM-P613 tablet (seed 42, 20 models per cycle, cycles 0-6). On CIFAR-10, cycle 1 is gate-accepted and improves the mobile deployment score approximately 25.6x over the baseline with a mean quantized accuracy of 46.9%; later cycles raise GPU accuracy but fail the non-decreasing mobile gate. On CIFAR-100, the pre-QLoRA baseline retains the best mobile score; iterative rounds improve GPU accuracy (up to 26.2%) yet cannot surpass cycle 0 on-device, and the training pool stalls at 19 examples after the first accepted round. Together, the two studies show that closed-loop GPU fine-tuning does not guarantee monotonic mobile gains, especially on harder classification tasks, and that multi-dataset, on-device measurement is needed to stress-test deployment objectives. We release per-cycle metrics with 95% confidence intervals, all figures, and complete reproduction commands.
Saif U Din, Muhammad Hussain, R. Timofte et al.· 0 citations
Large Mixture-of-Experts (MoE) language models are attractive for end-device deployment because only a small subset of experts is active per token, but their routed expert weights often exceed accelerator memory. We target latency-critical single-user settings where routed experts are staged on demand from CPU memory to a GPU or from Flash to a mobile NPU. In this setting, self-speculative decoding faces a new bottleneck: increasing the draft expert set improves accuracy but triggers extra expert loading, while cheap small-footprint drafts have low acceptance; moreover, verifying a multi-token block activates the union of target experts and is no longer close to one target step. We propose DraftExpert, an expansion-aware self-speculative decoding framework for expert-offloaded MoE inference. DraftExpert trains one lightweight accelerator-resident draft expert per layer by self-distilling residual, logit/token, and router-agreement signals from the frozen target MoE. At inference time, it uses a fixed-footprint shared+top-1+draft-expert drafter together with confidence--expansion truncation and target-expert prefetching, while final tokens are still exactly verified by the target model. On DeepSeek-V2-Lite and Moonlight-16B-A3B across CPU-GPU and Flash-NPU offload, DraftExpert improves decode throughput by 1.45x on average, raises draft acceptance to 84~87%, and achieves 86~88% prefetch hit rates.