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

 hjelp

-0.39

 selezione

-0.38

intervento

-0.38

 brukt

-0.35

 säga

-0.35

[]:

-0.34

 hjælp

-0.34

 denen

-0.34

Step

-0.34

 кӀ

-0.33

POSITIVE LOGITS

fasterxml

1.60

 BoxFit

0.69

jsdelivr

0.69

səhifə

0.65

AccessorTable

0.60

 Faster

0.58

الدراسه

0.57

fortawesome

0.56

jackson

0.54

flashdata

0.54

Activations Density 0.000%

No Known Activations

This feature has no known activations.