Long context fine-tuning of large language models(LLMs) involves training on datasets that are predominantly composed of short sequences and a small proportion of longer sequences. However, existing approaches overlook this long-tail distribution and employ training strategies designed specifically for long sequences. Moreover, these approaches also fail to address the challenges posed by variable sequence lengths during distributed training, such as load imbalance in data parallelism and severe pipeline bubbles in pipeline parallelism. These issues lead to suboptimal training performance and poor GPU resource utilization. To tackle these problems, we propose a chunk-centric training method named ChunkFlow. ChunkFlow reorganizes input sequences into uniformly sized chunks by consolidating short sequences and splitting longer ones. This approach achieves optimal computational efficiency and balance among training inputs. Additionally, ChunkFlow incorporates a state-aware chunk scheduling mechanism to ensure that the peak memory usage during training is primarily determined by the chunk size rather than the maximum sequence length in the dataset. Integrating this scheduling mechanism with existing pipeline scheduling algorithms further enhances the performance of distributed training. Experimental results demonstrate that, compared with Megatron-LM, ChunkFlow can be up to 4.53x faster in the long context fine-tuning of LLMs. Furthermore, we believe that ChunkFlow serves as an effective solution for a broader range of scenarios, such as long context continual pre-training, where datasets contain variable-length sequences.

Embedding columns are important for deep recommendation models to achieve high accuracy, but they can be very time-consuming during inference. Machine learning (ML) compilers are used broadly in real businesses to optimize ML models automatically. Unfortunately, no existing work uses compilers to automatically accelerate the heavy embedding column computations during recommendation model inferences. To fill this gap, we propose RECom, the first ML compiler that aims at optimizing the massive embedding columns in recommendation models on the GPU. RECom addresses three major challenges. First, generating an efficient schedule on the GPU for the massive operators within embedding columns is difficult. Existing solutions usually lead to numerous small kernels and also lack inter-subgraph parallelism. We adopt a novel codegen strategy that fuses massive embedding columns into a single kernel and maps each column into a separate thread block on the GPU. Second, the complex shape computations under dynamic shape scenarios impede further graph optimizations. We develop a symbolic expression-based module to reconstruct all shape computations. Third, ML frameworks inevitably introduce redundant computations due to robustness considerations. We develop a subgraph optimization module that performs graph-level simplifications based on the entire embedding column context. Experiments on both in-house and open-source models show that RECom can achieve 6.61X and 1.91X over state-of-the-art baselines in terms of end-to-end inference latency and throughput, respectively. RECom's source code is publicly available at https://github.com/AlibabaResearch/recom.

Compiler optimization plays an increasingly important role to boost the performance of machine learning models for data processing and management. With increasingly complex data, the dynamic tensor shape phenomenon emerges for ML models. However, existing ML compilers either can only handle static shape models or expose a series of performance problems for both operator fusion optimization and code generation in dynamic shape scenes. This paper tackles the main challenges of dynamic shape optimization: the fusion optimization without shape value, and code generation supporting arbitrary shapes. To tackle the fundamental challenge of the absence of shape values, it systematically abstracts and excavates the shape information and designs a cross-level symbolic shape representation. With the insight that what fusion optimization relies upon is tensor shape relationships between adjacent operators rather than exact shape values, it proposes the dynamic shape fusion approach based on shape information propagation. To generate code that adapts to arbitrary shapes efficiently, it proposes a compile-time and runtime combined code generation approach. Finally, it presents a complete optimization pipeline for dynamic shape models and implements an industrial-grade ML compiler, named BladeDISC. The extensive evaluation demonstrates that BladeDISC outperforms PyTorch, TorchScript, TVM, ONNX Runtime, XLA, Torch Inductor (dynamic shape), and TensorRT by up to 6.95×, 6.25×, 4.08×, 2.04×, 2.06×, 7.92×, and 4.16× (3.54×, 3.12×, 1.95×, 1.47×, 1.24×, 2.93×, and 1.46× on average) in terms of end-to-end inference speedup on the A10 and T4 GPU, respectively. BladeDISC's source code is publicly available at https://github.com/alibaba/BladeDISC.

With the fast growth of parameter size, it becomes increasingly challenging to deploy large generative models as they typically require large GPU memory consumption and massive computation. Unstructured model pruning has been a common approach to reduce both GPU memory footprint and the overall computation while retaining good model accuracy. However, the existing solutions do not provide an efficient support for handling unstructured sparsity on modern GPUs, especially on the highly-structured tensor core hardware. Therefore, we propose Flash-LLM for enabling low-cost and highly efficient large generative model inference with the sophisticated support of unstructured sparsity on high-performance but highly restrictive tensor cores. Based on our key observation that the main bottleneck of generative model inference is the several skinny matrix multiplications for which tensor cores would be significantly under-utilized due to low computational intensity, we propose a general Load-as-Sparse and Compute-as-Dense methodology for unstructured sparse matrix multiplication (SpMM). The basic insight is to address the significant memory bandwidth bottleneck while tolerating redundant computations that are not critical for end-to-end performance on tensor cores. Based on this, we design an effective software framework for tensor core based unstructured SpMM, leveraging on-chip resources for efficient sparse data extraction and computation/memory-access overlapping. Extensive evaluations demonstrate that (1) at SpMM kernel level, Flash-LLM significantly outperforms the state-of-the-art library, i.e., Sputnik and SparTA by an average of 2.9X and 1.5X, respectively.(2) At end-to-end framework level on OPT-30B/66B/175B models, for tokens per GPU-second , Flash-LLM achieves up to 3.8X and 3.6X improvement over DeepSpeed and FasterTransformer, respectively, with significantly lower inference cost.

As graph data is prevalent for an increasing number of Internet applications, continuously monitoring structural patterns in dynamic graphs in order to generate real-time alerts and trigger prompt actions becomes critical for many applications. In this paper, we present a new system GraphS to efficiently detect constrained cycles in a dynamic graph, which is changing constantly, and return the satisfying cycles in real-time. A hot point based index is built and efficiently maintained for each query so as to greatly speed-up query time and achieve high system throughput. The GraphS system is developed at Alibaba to actively monitor various online fraudulent activities based on cycle detection. For a dynamic graph with hundreds of millions of edges and vertices, the system is capable to cope with a peak rate of tens of thousands of edge updates per second and find all the cycles with predefined constraints with a 99.9% latency of 20 milliseconds.