INDEX

Explanations

code defining classes and class hierarchies

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

RegressionTest

-0.87

makeConstraints

-0.78

 betweenstory

-0.77

ftagPool

-0.74

 queſta

-0.73

ésultats

-0.72

évaluateur

-0.71

Spolja

-0.71

 مشين

-0.71

Hentet

-0.71

POSITIVE LOGITS

 hilo

0.35

special

0.28

for

0.27

↵↵

0.27

Den

0.27

子

0.26

An

0.26

 Stand

0.26

 with

0.26

Activations Density 0.004%

code defining classes and class hierarchies

No Comments

No Known Activations