INDEX

Explanations

phrases indicating influence or impact in various contexts

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

heavily influenced on

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

Diwedd

-0.54

UnitTesting

-0.51

AutoScaleMode

-0.50

 voeding

-0.49

⟬

-0.49

TextInputLayout

-0.49

WriteLiteral

-0.47

 triplicate

-0.47

 infinite

-0.47

まさかの

-0.47

POSITIVE LOGITS

 מאוד

0.62

Molto

0.51

 Veel

0.49

tagHelperRunner

0.48

 Many

0.47

Many

0.47

 heavily

0.46

Muito

0.46

 banget

0.45

 importants

0.44

Activations Density 0.088%

phrases indicating influence or impact in various contexts

heavily influenced on

No Comments

No Known Activations