INDEX

Explanations

keywords related to identification or naming

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hpur

-0.43

LookAnd

-0.42

hoga

-0.41

سطس

-0.40

fortawesome

-0.39

esp

-0.35

 odst

-0.35

 tempo

-0.34

beitet

-0.34

 rese

-0.33

POSITIVE LOGITS

 EconPapers

0.59

rungsseite

0.51

Cyfarwyddwr

0.50

 prefeitura

0.50

ValueStyle

0.49

Obrázky

0.49

 насељу

0.48

 للمعارف

0.47

 sorella

0.47

 ProtoMessage

0.46

Activations Density 0.000%

keywords related to identification or naming

No Comments

No Known Activations