INDEX

Explanations

terms related to finishing touches or final details

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

finish and its results

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

 vectorielle

-0.62

 wektor

-0.57

mparas

-0.57

 veckor

-0.54

 vectorielles

-0.54

Rodney

-0.53

 Thebes

-0.53

 NYPD

-0.52

trane

-0.52

ibol

-0.52

POSITIVE LOGITS

finish

1.04

 Finish

1.00

 finish

0.96

Finish

0.93

 FINISH

0.90

FINISH

0.90

 finishes

0.75

 finisher

0.74

 Finishing

0.74

 finishing

0.72

Activations Density 0.013%

terms related to finishing touches or final details

finish and its results

No Comments

No Known Activations