INDEX

Explanations

surprising or counterintuitive information

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

surprising, evidently, none

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

 NACIONAL

-0.43

OGND

-0.42

 pozi

-0.41

WillAppear

-0.41

loroethane

-0.39

ImageContext

-0.38

uhnya

-0.38

ORETICAL

-0.37

plotlib

-0.37

 Früchte

-0.37

POSITIVE LOGITS

 تضيفلها

0.57

 surprise

0.52

 shocking

0.50

surprise

0.47

CppMethod

0.47

 unsus

0.46

 surprising

0.46

 surprises

0.44

Autoritní

0.43

 shock

0.43

Activations Density 0.039%

surprising or counterintuitive information

surprising, evidently, none

No Comments

No Known Activations