INDEX

Explanations

phrases or concepts related to evaluating and improving processes or systems

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

following coordinating conjunction "and"

np_acts-logits-general · gemini-2.0-flash

action or state following 'and'

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

onload

-0.53

inen

-0.52

ariat

-0.50

fml

-0.50

ExecuteReader

-0.49

 نگاه

-0.49

 esquecer

-0.47

 glanced

-0.47

etal

-0.47

 visitare

-0.46

POSITIVE LOGITS

 accordingly

0.96

 thereupon

0.80

SharedDtor

0.77

 AssemblyTitle

0.76

 Accordingly

0.73

endphp

0.72

 then

0.72

titleMargin

0.71

 thence

0.68

 disambiguazione

0.67

Activations Density 0.386%

phrases or concepts related to evaluating and improving processes or systems

following coordinating conjunction "and"

action or state following 'and'

No Comments

No Known Activations