INDEX

Explanations

hyphenated words expressing negative attributes or excessiveness

oai_token-act-pair · gpt-3.5-turbo

negative phrases or themes related to oversimplification and exaggeration

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

ulhu

-0.97

 Dickinson

-0.92

 Burr

-0.87

 Juliet

-0.83

 Graveyard

-0.78

 Cassidy

-0.78

DeV

-0.74

 Boone

-0.74

 Blow

-0.74

Vid

-0.72

POSITIVE LOGITS

sized

1.25

regulation

1.11

regulated

1.09

region

1.07

spec

1.06

exc

1.06

emphasis

1.04

matched

1.03

reg

1.03

performing

1.03

Activations Density 0.046%

hyphenated words expressing negative attributes or excessiveness

negative phrases or themes related to oversimplification and exaggeration

No Comments

No Known Activations