INDEX

Explanations

terms related to interrogation processes and questioning techniques

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 للاسماء

-0.81

########.

-0.70

 BorderSide

-0.69

 дописавши

-0.66

 BoxDecoration

-0.66

thums

-0.66

bootstrapcdn

-0.64

 AssemblyCulture

-0.64

 surla

-0.63

⟬

-0.63

POSITIVE LOGITS

 interrog

0.84

 interrogation

0.80

 interrogated

0.74

 questions

0.65

 interro

0.62

 questioning

0.60

 query

0.55

 probing

0.54

 queries

0.54

 inquiry

0.54

Activations Density 0.022%

terms related to interrogation processes and questioning techniques

No Comments

No Known Activations