83 repos across 5 sub-areas
A broad collection of large language model repositories spanning various model architectures, sizes, and implementations. The cluster includes popular open-source models like Qwen, GLM, MiniCPM, and others, with significant representation of on-device and edge-optimized variants. While the dominant languages (Swift, Python) and topics (apple-silicon, macos) suggest a secondary focus on deploying these models on Apple platforms, the cluster's core is fundamentally about LLM implementations, training frameworks, and model artifacts across the wider ecosystem.
On-Device Speech & Audio AI Models
25 repos
Lightweight, optimized speech recognition, text-to-speech, and voice processing models designed for on-device inference on mobile and edge hardware. This cluster focuses on compact model architectures (typically under 2B parameters) that enable real-time audio applications without cloud dependencies, featuring streaming ASR systems, voice synthesis, and multimodal voice models from vendors like NVIDIA, Alibaba, and Apple-compatible platforms.
Cluster 650220
24 repos
Cluster 650219
22 repos
Cluster 650218
10 repos
Core AI Model Implementations
2 repos
Optimized implementations and variants of large language models and multimodal models designed for efficient deployment, particularly on resource-constrained or specialized hardware. The cluster centers on quantized and distilled versions of popular LLM architectures (MiniCPM, Qwen, Gemma, Nanbeige) alongside platform-specific integrations—notably Apple's Core AI framework for on-device inference on iOS and macOS. These repositories reflect the practical engineering of making state-of-the-art models deployable in production environments with strict latency and memory constraints.