INDEX

Explanations

temporal phrases related to the effects and outcomes of treatments

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

outcomes after action

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

 itſelf

-0.80

ագրություններ

-0.71

 fashiola

-0.70

 queſto

-0.68

ⓧ

-0.67

𑄧

-0.63

𑄟

-0.63

๣

-0.63

 propOrder

-0.63

AndEndTag

-0.63

POSITIVE LOGITS

ďaka

0.30

 this

0.28

 jspb

0.26

 dzięki

0.26

 nhờ

0.25

拠

0.25

 efforts

0.24

 esforço

0.24

новништво

0.23

 Rich

0.23

Activations Density 0.025%

temporal phrases related to the effects and outcomes of treatments

outcomes after action

No Comments

No Known Activations