INDEX

Explanations

phrases related to accuracy and reliability of information

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

warranties and disclaimers

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

principalColumn

-0.48

 fusca

-0.47

 InputDecoration

-0.45

zones

-0.43

 Վերցված

-0.40

Gemeinden

-0.40

 wireType

-0.39

 niyang

-0.38

shifting

-0.38

otry

-0.38

POSITIVE LOGITS

ThroughAttribute

0.52

 cellspacing

0.48

 transfieras

0.46

 defStyle

0.44

httphttps

0.43

xase

0.41

 siquiera

0.41

 joão

0.41

 quelconque

0.41

цезда

0.40

Activations Density 0.017%

phrases related to accuracy and reliability of information

warranties and disclaimers

No Comments

No Known Activations