INDEX

Explanations

specific identifiers and references within various contexts, including people, objects, and categories

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-0.33

[toxicity=0]

-0.32

 repetir

-0.32

 Verarbeitung

-0.31

 hatır

-0.31

 käs

-0.31

extérieur

-0.30

 JADX

-0.30

-0.29

POSITIVE LOGITS

Personensuche

0.84



0.66

 ब्रेकडाउन

0.66

InjectAttribute

0.64

 ModelExpression

0.63

Diweddarwch

0.62

IntoConstraints

0.61

ſcher

0.60

 tartalomajánló

0.60

Tikang

0.59

Activations Density 0.104%

specific identifiers and references within various contexts, including people, objects, and categories

No Comments

No Known Activations