INDEX

Explanations

individuals who have spoken publicly about a variety of topics

oai_token-act-pair · gpt-3.5-turbo

instances of the word "spoke" to indicate communication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

\/\/

-0.72

assi

-0.70

aredevil

-0.68

 Ryder

-0.66

rawler

-0.66

mental

-0.66

=~

-0.65

xon

-0.65

vernment

-0.64

@#&

-0.64

POSITIVE LOGITS

 volumes

0.92

 aloud

0.83

 transcripts

0.81

 fluent

0.76

 Mandarin

0.76

 mute

0.75

 louder

0.75

 spoken

0.74

 voices

0.73

 Cosponsors

0.73

Activations Density 0.029%

individuals who have spoken publicly about a variety of topics

instances of the word "spoke" to indicate communication

No Comments

No Known Activations