INDEX

Explanations

references to pronouns and their associated subjects

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

})*/

-0.52

rovna

-0.46

 classed

-0.45

efois

-0.44

ciclop

-0.43

engesch

-0.42

colgante

-0.42

zyl

-0.42

});*/

-0.42

 histamine

-0.41

POSITIVE LOGITS

complexContent

0.48

évaluateur

0.46

 consecuencia

0.44

Vidite

0.44

DataAnnotations

0.42

 expectations

0.42

featureID

0.42

AISSEE

0.40

ServletException

0.40

vrigt

0.40

Activations Density 0.075%

references to pronouns and their associated subjects

No Comments

No Known Activations