INDEX

Explanations

phrases that indicate specific attributes or conditions related to requests

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

UserScript

-0.52

entur

-0.45

seeded

-0.45

ponses

-0.44

 Rast

-0.44

takan

-0.43

gsSP

-0.43

 DataTypes

-0.43

esinde

-0.42

verket

-0.41

POSITIVE LOGITS

 vilket

0.66

 Which

0.63

 WHICH

0.62

 mikä

0.60

Which

0.59

 which

0.58

which

0.56

IntoConstraints

0.53

 hvilket

0.51

 vilken

0.49

Activations Density 0.365%

phrases that indicate specific attributes or conditions related to requests

No Comments

No Known Activations