INDEX

Explanations

negative prefixes, particularly "un-"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 1-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_1/width_16k/average_l0_102

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.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

 pleaſure

-1.01

 ſeveral

-1.00

 itſelf

-0.90

 Monfieur

-0.90

 Chriftian

-0.89

 becauſe

-0.87

 leaſt

-0.85

 Diſ

-0.85

Efq

-0.83

 Jefus

-0.83

POSITIVE LOGITS

un

1.75

pe

0.89

featureID

0.89

sub

0.89

Un

0.86

dis

0.86

una

0.85

cre

0.81

 blind

0.78

ex

0.78

Activations Density 0.113%

negative prefixes, particularly "un-"

No Comments

No Known Activations