INDEX

Explanations

phrases that indicate contrast or objection

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

but followed by qualification

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-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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 unknownFields

-0.56

 TextInputType

-0.55

InputBorder

-0.54

 wireType

-0.53

featureID

-0.52

MemoryWarning

-0.52

majánló

-0.52

 queſta

-0.51

 initComponents

-0.51

uxxxx

-0.51

POSITIVE LOGITS

 consultato

0.36

arraycopy

0.33

but

0.32

ftagPool

0.32

LookAnd

0.32

зулта

0.31

EnableWeb

0.31

AddHtmlAttribute

0.31

But

0.31

 enough

0.30

Activations Density 0.038%

phrases that indicate contrast or objection

but followed by qualification

No Comments

No Known Activations