hyperquader/ministral-3-3b-instruct-genai-int4
Updated
Proof of concept for integrating AI capabilities into the Linux Mint Cinnamon desktop.
Note Format conversion of Ministral 3 3B Instruct to an onnxruntime-genai int4 directory. Not a fine-tune.
Note A format conversion, not a fine-tune: Ministral 3 8B Instruct as an ONNX Runtime GenAI int4 directory, CUDA only.
Note Fine-tune: merged bf16 weights, Amberlin's tool calls and the answers after them.
Note Fine-tune: the LoRA adapter alone.
Note Fine-tune: the int4 CUDA build the broker loads.
Note Fine-tune: merged bf16 weights, the CPU tier's tool calls and the answers after them.
Note Fine-tune: the LoRA adapter alone.
Note Fine-tune: the int4 CPU and CUDA builds the broker loads.