INDEX

Explanations

phrases emphasizing the significance of specific subjects or concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

testens

-0.44

OrWhiteSpace

-0.40

 hang

-0.37

avadoc

-0.37

envolvimento

-0.37

Adapun

-0.36

 sacrific

-0.35

 chaff

-0.35

 behind

-0.35

 hangs

-0.34

POSITIVE LOGITS

 CanadaChoose

0.65

 important

0.60

 beginnetje

0.58

 autorytatywna

0.56

important

0.52

subpackage

0.50

 worthwhile

0.49

 phénomènes

0.47

importante

0.47

 belangrijke

0.47

Activations Density 0.023%

phrases emphasizing the significance of specific subjects or concepts

No Comments

No Known Activations