Extremely compressed LLMs for deployment on edge devices
AI & ML interests
None defined yet.
Recent Activity
Organization Card
TheStage AI is a team of AI researchers and engineers focused on efficient model inference.
On Hugging Face, we publish Elastic checkpoints (Transformers, Diffusers, ASR) optimized for fast, cost-efficient serving.
Elastic Models are released as ready-to-run checkpoints and performance tiers (S / M / L / XL) so you can choose the best balance of quality, latency, and memory for your workload across NVIDIA GPUs (incl. Jetson), Apple Silicon, and edge devices.
Pipeline (what makes these HF releases): ANNA (analyzes constraints & targets) • Qlip (quantizes/sparsifies + compiles for optimized runtime & serving) • CLI/Platform (run jobs, deploy, benchmark).
Community: X • Discord (request invite) • YouTube • LinkedIn
models 73
TheStageAI/embeddinggemma-300m
Updated • 84
TheStageAI/gemma-4-12B-it-GGUF
Text Generation • 12B • Updated • 1.14k • 4
TheStageAI/gemma-4-E4B-it-GGUF
Text Generation • 7B • Updated • 852 • 3
TheStageAI/gemma-4-E2B-it-GGUF
Text Generation • 5B • Updated • 440 • 1
TheStageAI/Qwen3.5-9B-GGUF
Text Generation • 9B • Updated • 382 • 1
TheStageAI/Qwen3.5-4B-GGUF
Text Generation • 4B • Updated • 412 • 1
TheStageAI/Qwen3.5-2B-GGUF
Text Generation • 2B • Updated • 526 • 1
TheStageAI/Qwen3.5-0.8B-GGUF
Text Generation • 0.8B • Updated • 310 • 1
TheStageAI/neutts-2e-0.2b
Updated
TheStageAI/smart-turn-v3
Updated
datasets 0
None public yet