INDEX

Explanations

references to coding assistance or helper functions

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

]")]

-0.56

tableFuture

-0.56

oplayer

-0.49

UrlResolution

-0.48

))->

-0.48

 kaarangay

-0.47

новништво

-0.47

afone

-0.45

DrawerToggle

-0.45

➌

-0.45

POSITIVE LOGITS

helper

0.96

RSpec

0.73

Helper

0.66

ViewFeatures

0.64

 helper

0.60

 Helper

0.59

IUrlHelper

0.52

 TestBed

0.52

 avoient

0.52

UnitTesting

0.49

Activations Density 0.000%

references to coding assistance or helper functions

No Comments

No Known Activations