INDEX

Explanations

phrases related to thematic or conceptual unity

oai_token-act-pair · gpt-3.5-turbo

references to a specific individual or their ideas

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

GROUND

-0.86

dated

-0.78

 stakes

-0.72

breaks

-0.72

 hazard

-0.71

 toilets

-0.70

 lawy

-0.64

 hazards

-0.64

 kinderg

-0.64

 inhibitors

-0.62

POSITIVE LOGITS

mmm

1.20

useum

1.05

achine

0.95

ilit

0.93

ajor

0.92

ature

0.92

agogue

0.91

oths

0.91

obile

0.90

otor

0.90

Activations Density 0.021%

phrases related to thematic or conceptual unity

references to a specific individual or their ideas

No Comments

No Known Activations