INDEX

Explanations

phrases related to writing or reviewing various forms of content

oai_token-act-pair · gpt-3.5-turbo

references to various types of written content or assignments

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

ItemThumbnailImage

-0.74

 vantage

-0.73

seekers

-0.72

elight

-0.68

Ability

-0.66

FML

-0.66

gap

-0.65

hunter

-0.65

FACE

-0.65

 locate

-0.62

POSITIVE LOGITS

 journal

0.70

 screenplay

0.69

obook

0.69

 Frankenstein

0.68

batch

0.68

 aloud

0.66

 journals

0.65

 autobiography

0.65

olkien

0.64

RW

0.62

Activations Density 0.160%

phrases related to writing or reviewing various forms of content

references to various types of written content or assignments

No Comments

No Known Activations