INDEX

Explanations

information related to television show airings

oai_token-act-pair · gpt-3.5-turbo

instances of the word "air" related to media programming schedules

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

usalem

-0.74

ledge

-0.68

usra

-0.67

nton

-0.66

 disciplined

-0.66

 prag

-0.66

cius

-0.65

hedral

-0.64

ourke

-0.64

orum

-0.63

POSITIVE LOGITS

waves

1.11

 airing

0.92

 airs

0.90

ONY

0.90

Ð´

0.90

 broadcasts

0.87

bender

0.86

 Sundays

0.86

 Seasons

0.85

 microphones

0.83

Activations Density 0.031%

information related to television show airings

instances of the word "air" related to media programming schedules

No Comments

No Known Activations