INDEX

Explanations

terms related to hidden or latent concepts

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 36-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_36/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.36.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

etcode

-0.38

 cuisson

-0.36

kháu

-0.35

 Italijanski

-0.33

 rapides

-0.32

 detenido

-0.31

 declaraciones

-0.31

Ordre

-0.31

 AssemblyProduct

-0.31

 atención

-0.31

POSITIVE LOGITS

 Hide

0.80

 hide

0.77

 doInBackground

0.75

 hiding

0.70

Hide

0.70

hid

0.70

 hides

0.69

 concealing

0.68

 esconde

0.67

Hid

0.67

Activations Density 0.166%

terms related to hidden or latent concepts

No Known Activations