INDEX

Explanations

references to physical barriers or structures, particularly walls

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

climbing on walls

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 31-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_31/width_131k/average_l0_109

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.31.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

 EconPapers

-0.54

 resourceCulture

-0.49

 betweenstory

-0.48

EndContext

-0.47

 الاطلاع

-0.45

RenderAtEndOf

-0.45

PerformLayout

-0.42

 ujednoznacz

-0.41

AddressBook

-0.41

 autorytatywna

-0.40

POSITIVE LOGITS

 walls

2.48

 wall

2.47

wall

2.06

walls

2.03

 Walls

1.96

 Wall

1.91

Wall

1.91

 WALL

1.88

 dinding

1.87

Walls

1.86

Activations Density 0.110%

references to physical barriers or structures, particularly walls

climbing on walls

No Comments

No Known Activations