Inference & Serving
Practical run logs, evaluation methods, build reports, and postmortems from LoRADock contributors.
Inference & Serving
5 articlesServing Four LoRA Adapters from One Base Model
A practical LoRADock note covering inputs, validation, tradeoffs, and release criteria for serving four lora adapters from one base model.
Safetensors or GGUF: What Worked Better on My Small Server
A practical LoRADock note covering inputs, validation, tradeoffs, and release criteria for safetensors or gguf: what worked better on my small server.
Reducing Cold-Start Time for Frequently Swapped Adapters
A practical LoRADock note covering inputs, validation, tradeoffs, and release criteria for reducing cold-start time for frequently swapped adapters.
Running a Quantized Text Adapter on a 16 GB Machine
A practical LoRADock note covering inputs, validation, tradeoffs, and release criteria for running a quantized text adapter on a 16 gb machine.
Tracking Memory Use During Concurrent Generation Requests
A practical LoRADock note covering inputs, validation, tradeoffs, and release criteria for tracking memory use during concurrent generation requests.