A CausalLM with bidirectional attention (an A2D diffusion model, DIFF-4) as a StepRelevanceModel, with the same rules as EncoderLMStepModel.
More...
#include <masked_diffusion_lrp.hpp>
A CausalLM with bidirectional attention (an A2D diffusion model, DIFF-4) as a StepRelevanceModel, with the same rules as EncoderLMStepModel.
◆ CausalLMStepModel()
- Parameters
-
| backend | Where the model runs. The model and backend must outlive this. |
- Exceptions
-
| std::invalid_argument | if the model is causal. |
◆ forward()
| std::vector< float > pulsatrix::CausalLMStepModel::forward |
( |
const std::vector< int64_t > & |
tokens | ) |
|
|
overridevirtual |
◆ relevance()
| std::vector< float > pulsatrix::CausalLMStepModel::relevance |
( |
const std::vector< float > & |
relevance_logits | ) |
|
|
overridevirtual |
The documentation for this class was generated from the following file: