Property Logits

Namespace

LMKit.TextGeneration.Events

Assembly

LM-Kit.NET.dll

Logits

An array of logit values representing the unnormalized log probabilities of all tokens.

public float[] Logits { get; }

Property Value

float[]

Remarks

Logits are the raw outputs of the last layer in a neural network before applying the softmax function to convert them into probabilities. Each element in this array corresponds to a token in the model's vocabulary, and the value represents the model's confidence in that token's relevance or likelihood in the given context. Accessing this array allows for a deeper analysis of the model's prediction behavior, enabling custom post-processing or interpretation of the predicted token distribution.

At this stage, the entries within the array are editable, permitting the introduction of direct biases.