INDEX

Explanations

terms related to publishing or publication sources

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

Publisher's note or information

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

ftagPool

-0.52

parametrize

-0.47

 AssemblyCulture

-0.47

balleur

-0.46

vician

-0.41

 يتيمه

-0.39

ysql

-0.39

Gön

-0.39

ActionCreators

-0.39

ChrTalk

-0.39

POSITIVE LOGITS

Publisher

0.85

publisher

0.81

 Publisher

0.77

 publisher

0.74

 publishers

0.60

featureID

0.57

 <<<<<<<<<<<<<<

0.52

builder

0.48

 Publishers

0.48

RTLI

0.48

Activations Density 0.001%

terms related to publishing or publication sources

Publisher's note or information

No Comments

No Known Activations