INDEX

Explanations

phrases indicating influence or guidance in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

Preceding words ending with "-ed" followed by "by"

np_acts-logits-general · gemini-2.0-flash

influenced or caused by

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 endfor

-0.40

 Seaside

-0.37

displaystyle

-0.37

currentTarget

-0.36

Hire

-0.35

 مف

-0.34

 Staat

-0.34

 Cape

-0.34

င်း

-0.34

drawLine

-0.33

POSITIVE LOGITS

 propOrder

0.81

bootstrapcdn

0.72

 resourceCulture

0.66

0.61

fromnode

0.59

 oleh

0.59

+#+#

0.59

 ModelRenderer

0.59

 tartalomajánló

0.58

 MainAxisSize

0.55

Activations Density 0.449%

phrases indicating influence or guidance in various contexts

Preceding words ending with "-ed" followed by "by"

influenced or caused by

No Comments

No Known Activations