INDEX

Explanations

No Explanations Found

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

IBOutlet

-0.64

 lenker

-0.53

InputTagHelper

-0.48

liferay

-0.47

ContentAlignment

-0.45

 Helfer

-0.45

 Söder

-0.45

PostInfinity

-0.44

onedDateTime

-0.44

Ape

-0.44

POSITIVE LOGITS

 Truy

4.13

 truy

1.33

 truyền

1.19

Tru

0.88

Tru

0.81

TRANSM

0.75

principalTable

0.75

傳

0.74

传

0.74

 Transmission

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.