INDEX

Explanations

headings or section titles within the text

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

Rhestr

-0.49

Personendaten

-0.44

IGraphics

-0.43

 protoimpl

-0.42

thenReturn

-0.41

onAttach

-0.40

andReturn

-0.38

 ComVisible

-0.37

 Wikimedijinoj

-0.37

 allemaal

-0.37

POSITIVE LOGITS

bibfield

0.63

BibitemShut

0.62

 queſta

0.57

 groupBox

0.57

 UserId

0.56

bibinfo

0.56

 pilar

0.55

 zwiſchen

0.54

motif

0.51

 zuſammen

0.51

Activations Density 0.000%

headings or section titles within the text

No Comments

No Known Activations