INDEX

Explanations

terms that describe size and magnitude

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

large followed by noun

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

-0.34

-0.30

 explain

-0.29

 perbuatan

-0.28

based

-0.28

 elkaar

-0.28

 jums

-0.28

POSITIVE LOGITS

IBOutlet

0.78

 surla

0.74

 imagui

0.73



0.72

 zwiſchen

0.72

ſſung

0.72

<unused42>

0.71

<unused41>

0.71

<pad>

0.71

[@BOS@]

0.71

Activations Density 0.046%

terms that describe size and magnitude

large followed by noun

No Comments

No Known Activations