INDEX

Explanations

negative descriptors or critical evaluations

oai_token-act-pair · gpt-3.5-turbo

references to traditional or old-fashioned concepts and terms

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

 Dickinson

-0.87

Wiz

-0.85

 Norris

-0.81

 Rica

-0.77

 Shutterstock

-0.74

 Cohn

-0.73

ulhu

-0.71

 Noir

-0.70

 Harvey

-0.69

HH

-0.69

POSITIVE LOGITS

sounding

1.33

minded

1.17

looking

1.16

advertisement

1.13

based

1.13

bodied

1.13

themed

1.12

sized

1.12

advertising

1.11

style

1.10

Activations Density 0.198%

negative descriptors or critical evaluations

references to traditional or old-fashioned concepts and terms

No Comments

No Known Activations