Awesome GitHub RepositoriesLong-Context Sequence Processors

Qwen3-Coder is a specialized large language model designed for software development, technical reasoning, and automated code synthesis. Built on transformer-based sequence modeling, it functions as a multilingual programming assistant capable of generating, completing, and debugging source code across more than one hundred programming languages. The model distinguishes itself through its capacity to process and maintain logical coherence across massive datasets, supporting context windows of up to one million tokens. This allows for repository-scale reasoning, enabling the model to analyze co

TensorRT-LLM is a platform and toolkit designed for compiling, optimizing, and serving transformer-based models on accelerated hardware. It functions as a framework that transforms machine learning models into efficient execution graphs, providing an engine to refine these models for specific hardware to maximize throughput and minimize latency during text generation. The project distinguishes itself through advanced execution strategies that manage the entire inference pipeline. It utilizes kernel-level fusion and static graph execution to optimize mathematical operations and computational f

Airllm is a framework designed to execute and fine-tune large language models on consumer-grade hardware. By employing layer-wise model decomposition and memory-efficient loading techniques, the engine enables the operation of massive models that would otherwise exceed available system or video memory. The project distinguishes itself through a suite of optimization strategies that balance memory footprint with performance. It utilizes block-wise weight quantization and asynchronous layer prefetching to reduce resource consumption and hide data transfer latency. Additionally, the framework su

Lingbot-map is a feed-forward neural network designed for real-time 3D scene reconstruction from streaming video. It processes video frames one at a time without iterative optimization, producing dense geometry and camera poses at interactive frame rates directly from a live feed. The project distinguishes itself through its ability to maintain stable geometry and pose alignment across very long video sequences, handling thousands of frames without drift. It achieves this through a combination of coordinate grounding memory, sliding-window inference with overlapping keyframes, and a paged KV

bert4keras 是 Keras 深度学习框架的 BERT Transformer 架构的轻量级重新实现。它作为一个自然语言处理工具包和 Transformer 模型库，用于文本分类、序列标注和语义嵌入提取。 该框架包括用于问答和文本生成的序列到序列模型系统，以及一个模型推理服务器，用于将训练好的 Transformer 部署为 Web API 以进行实时预测。 功能涵盖了广泛的自然语言理解任务，包括阅读理解、关系提取和长文本处理。该库提供了用于语言模型预训练和微调的工具，以及参数缩减、用于鲁棒性的对抗性训练和层级学习率配置等优化技术。 该项目包括一个权重转换加载器，用于将外部格式的预训练权重转换为兼容的 Keras 结构。

Flash Linear Attention is a training framework and inference engine for sequence models that use linear attention and state space mechanisms, designed to process long contexts with reduced memory and compute overhead. It provides hardware-optimized token mixing layers and fused CUDA kernels that minimize memory bandwidth and launch overhead across different GPU architectures, and includes a causal inference engine that generates text token-by-token using cached hidden states for efficient autoregressive decoding. The project supports building hybrid sequence models that interleave standard at

DeepSeek-V2 是一款专为自然语言处理和长文本序列分析设计的大语言模型。它采用混合专家（MoE）架构，在高性能与推理效率之间取得了平衡。 该模型利用稀疏路由机制和共享专家神经元来获取通用知识，同时保持专业化能力。它通过多头潜在注意力（MLA）、分组查询注意力（GQA）和低秩张量压缩，进一步降低了内存开销并提高了吞吐量。 这些能力使其能够处理和检索海量 Token 信息，并通过降低硬件成本和内存瓶颈支持经济高效的部署。该系统兼容标准 API 接口，可与现有的语言模型工具链集成。

mini-sglang is a collection of tools for large language model inference, serving as an OpenAI-compatible inference server, a memory-efficient prefill engine, and a tensor parallelism runtime. It also functions as a local batch processing engine for offline benchmarking and ablation studies. The project focuses on acceleration and memory management through a KV cache manager that reuses precomputed caches for shared request prefixes. It handles large model workloads by distributing tasks across multiple GPUs and manages peak memory consumption by splitting long input sequences into smaller chu