INDEX

Explanations

concepts related to the evaluation and purpose of actions or entities

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

 препратки

-0.61

 Picchu

-0.57

󠁮

-0.56

ScopeManager

-0.52

 TextAlign

-0.51

 Administrativna

-0.51

orteur

-0.49

}{@

-0.49

initWithFrame

-0.49

iſen

-0.48

POSITIVE LOGITS

DebuggerStep

0.38

 avis

0.38

 Polisi

0.35

 psychologie

0.35

 politie

0.35

 Leser

0.35

jeu

0.34

 Constitucional

0.34

 gants

0.34

unehmen

0.34

Activations Density 0.045%

concepts related to the evaluation and purpose of actions or entities

No Comments

No Known Activations