INDEX

Explanations

phrases relating to the analysis and evaluation of evidence and arguments

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

explaining why

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ſicht

-0.85

niſſe

-0.79

 geſch

-0.79

ðsíða

-0.79

iſchen

-0.77

 témoig

-0.77

majánló

-0.74

 Geiſt

-0.74

 queſto

-0.74

ſſung

-0.73

POSITIVE LOGITS

our

0.28

are

0.28

Tre

0.24

the

0.24

 cons

0.24

et

0.24

 united

0.24

0.23

Con

0.23

AnchorStyles

0.23

Activations Density 0.124%

phrases relating to the analysis and evaluation of evidence and arguments

explaining why

No Comments

No Known Activations