implement-moe-mamba-model-for-text-generation
Implement a PyTorch-based MoE-Mamba model featuring an input-dependent selection mechanism and Mixture of Experts (MoE) layer for text generation tasks, including data loading, training, and evaluation workflows.
c-linux-socket-wrapper-implementation
Implement a C++ wrapper for Linux TCP sockets and file descriptors with unique ownership semantics, including FileDescriptor, Socket, Connection, Server, and Client classes.
c-tcp-socket-optimization-function
Generates a C function to configure TCP socket options (TCP_NODELAY, TCP_CORK, TCP_NOPUSH, TCP_QUICKACK, IP_TOS) for either low latency or high throughput using traditional if statements.
pytorch-nnparameter
Создание пользовательского слоя nn.Module, веса которого являются обучаемыми параметрами (torch.nn.Parameter) и обновляются через loss.backward() и optimizer.step(), вместо статического вычисления на входе.
pytorch-character-level-transformer-with-8-bit-vocabulary
Implement a PyTorch Transformer model using nn.Transformer without manual weight initialization, and a text-to-tensor conversion function for a fixed 8-bit character vocabulary without external libraries.
pytorch-character-level-text-to-tensor-conversion
Converts a raw string into a PyTorch tensor of indices using a fixed 8-bit character vocabulary, without external libraries, suitable for input into an embedding layer.
c-bigint-class-with-reverse-vector-storage
Implement a C++ BigInt class for arbitrarily large integers using a vector of digits stored in reverse order, supporting string conversion and addition.
python-oop-senior-level-interview-trainer
Provides concise, bullet-pointed explanations of technical concepts (specifically OOP) tailored for senior engineers, emphasizing Python-specific implementations and nuances for interview preparation.
gnn-edge-feature-embedding-generation
Generates PyTorch embeddings for categorical edge features in a Graph Neural Network by mapping string values to indices and concatenating learned embeddings according to a specific structure.
memento-customvector-cow-cpp
Реализовать классы Memento и CustomVector на C++ с использованием паттерна Copy-on-Write (COW) для эффективного управления памятью и валидности снимков.
freertos-64bit-division-implementation
Implements 64-bit division functions (div_u64, div_s64, div64_u64, div64_s64) for FreeRTOS on 32-bit architectures using bitwise shifting algorithms, incorporating specific sign handling and quotient adjustment logic.
generate-aspnet-mvc-entity-framework-models-from-database-schema
Generates C# POCO classes for Entity Framework based on a database schema, applying specific data annotations for keys, unique constraints, foreign keys, and data types like byte arrays for passwords.
gnn-edge-feature-embedding-generator
Generates PyTorch embeddings for graph edge features by mapping categorical strings to indices and concatenating learned embeddings, specifically handling device, net, and terminal attributes.
windows-process-memory-manipulation-client
Generates C++ client code to find process IDs, module base addresses, and read/write memory via a custom kernel driver or Windows API, ensuring correct syntax and error handling.
u-net-pytorch
Создание класса U-Net и функций encoder_block/decoder_block в PyTorch, соответствующих конкретным требованиям к структуре слоев (Conv->ReLU->Pool/Upsample) и логике skip-connections, чтобы пройти заданные проверки (assertions).
c-enable-if
根据用户指定的条件（如指针引用、继承关系），使用 std::enable_if 限制 C++ 模板函数的实例化，并移除不符合条件的重载。
keraslstmcpu
构建字符级LSTM模型进行文本生成，解决Tokenizer索引越界问题，并配置CPU多进程训练优化。
cmapunorderedmapallocator
编写C++通用模板函数，支持std::map和std::unordered_map作为通用容器传入，处理const正确性，支持可选的allocator模板参数，且禁止在函数内部构建临时对象。
pytorch-fusedbun-optimizer-implementation
Generates a new PyTorch optimizer class by fusing logic from two provided source implementations. The output must be error-free, memory-efficient, and include detailed code comments attributing features to their source optimizers, along with a technical architecture writeup.
pytorch-rnn-dataset-chunking-configuration
Modifies the data preparation phase of a PyTorch RNN/LSTM training script to limit the dataset size by dividing it into chunks. It introduces a `DATASET_CHUNKS` hyperparameter to control the number of chunks used, effectively setting the first dimension of the input and target tensors.