coreai_opt.quantization.spec.StatelessQParamsCalculatorBase

class coreai_opt.quantization.spec.StatelessQParamsCalculatorBase(**kwargs)

Bases: QParamsCalculatorBase

Stateless base: no cached qparams; recomputed every forward.

Used for dynamic quantization where activations vary per inference and the scale shape may change across forwards (e.g. LLM token-wise with variable sequence length). self.scale/zero_point/minval are assigned in forward as plain Python attributes (not buffers) for debugging visibility only — they reflect the most recent forward and are not in state_dict.

• FakeQuantizeImplBase keeps observer_enabled = 1 for this subclass so the recompute path stays live.

• get_qparams is undefined; set_export_mode(True) raises; float_range=[None, None] is required.

__init__(**kwargs)

Methods

compute_qparams(tensor, min_val, max_val)	Given the observed min/max range, return (scale, zero_point, minval).
forward(tensor)	Compute qparams from tensor and return; no buffer state.
get_class(key)
list_registry_keys()
list_registry_values()
register(key)	Register a virtual subclass of an ABC.
resolve(data)	Resolve a string key or class type against this registry.
set_export_mode([enabled])

forward(tensor)

Compute qparams from tensor and return; no buffer state.

Parameters:

tensor (Tensor)

Return type:

tuple[Tensor, Tensor | None, Tensor | None]

set_export_mode(enabled=True)

Parameters:

enabled (bool)

Return type:

None