INDEX

Explanations

symbols and formatting related to lists and itemization

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

 dieß

-0.75

 EconPapers

-0.75

AndEndTag

-0.74

niſſe

-0.74

iſchen

-0.74

SequentialGroup

-0.73

 ſind

-0.71

 faſt

-0.71

tagHelperRunner

-0.70



-0.69

POSITIVE LOGITS

The

0.58

The

0.46

TheReal

0.43

0.42

the

0.42

《

0.39

Read

0.39

On

0.37

Star

0.36

Activations Density 0.014%

symbols and formatting related to lists and itemization

No Comments

No Known Activations