INDEX

Explanations

elements related to interviews and public statements

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

AndEndTag

-0.74

 <>",

-0.56

 AssemblyCulture

-0.56

TagMode

-0.50

 démocr

-0.47

 Romains

-0.46

 للمعارف

-0.45

 pauvres

-0.44

jsxFileName

-0.43

ksikon

-0.43

POSITIVE LOGITS

 radio

0.44

 telegram

0.42

 tweeted

0.41

 publi

0.39

 undec

0.39

fabricated

0.38

+#+

0.38

 crack

0.38

 lobby

0.38

 sling

0.38

Activations Density 0.016%

elements related to interviews and public statements

No Comments

No Known Activations