INDEX

Explanations

references to study goals and outcomes

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

groups, days, weeks, concepts, fields, Nos

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

 exemption

-0.40

 stimulation

-0.39

wijs

-0.39

 payment

-0.39

 another

-0.38

 perforation

-0.38

insatz

-0.37

ίκη

-0.37

payment

-0.36

 Stimulation

-0.36

POSITIVE LOGITS

分别是

0.74

::{

0.70

AddTagHelper

0.68

InitVars

0.68

LookAnd

0.65

DoubleQuotes

0.64

 BoxDecoration

0.63

 nahilalakip

0.62

 Messrs

0.59

Drs

0.58

Activations Density 0.200%

references to study goals and outcomes

groups, days, weeks, concepts, fields, Nos

No Comments

No Known Activations