INDEX

Explanations

conjunctions and qualifiers related to research design and outcomes

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

hoeddwyd

-0.46

ביוגרפיה

-0.46

SceneManagement

-0.45

dersfield

-0.44

 Diag

-0.44

arakhand

-0.43

<bos>

-0.43

 abortions

-0.43

削

-0.42

domada

-0.42

POSITIVE LOGITS

 argint

0.48

yfik

0.42

 везде

0.41

 everywhere

0.40

UnsafeEnabled

0.38

 NSCoder

0.38

 Einrichtungen

0.37

 jedes

0.37

Datuak

0.37

 ovunque

0.37

Activations Density 0.201%

conjunctions and qualifiers related to research design and outcomes

No Comments

No Known Activations