Quant.h File
Quant Node — CVU kernel that quantizes FP32 to INT8 (scale + zero-point). More...
Included Headers
#include "builder/NodeContractConfigurable.h"
#include "builder/NodeContractProvider.h"
#include "builder/Node.h"
#include <nlohmann/json.hpp>
#include <memory>
#include <optional>
#include <string>
#include <vector>
Namespaces Index
| namespace | simaai |
| namespace | neat |
| namespace | nodes |
Classes Index
| struct | QuantOptions |
| class | Quant |
|
CVU kernel Node that quantizes FP32 tensors to INT8 using scale + zero-point. More... | |
Description
Quant Node — CVU kernel that quantizes FP32 to INT8 (scale + zero-point).
Reads FP32 tensors and emits INT8 using the per-tensor scale and zero-point bound to the model. Inserted at the MLA boundary on the INT8 path before the MLA when the upstream stage emits FP32.
- See Also
"The dtype contract" page in /concepts/dtype_contract
Generated via doxygen2docusaurus 2.0.0 by Doxygen 1.9.8.