INDEX

Explanations

proper nouns, particularly from the same source

oai_token-act-pair · gpt-3.5-turbo

mentions of the media outlet "Vox"

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

brow

-0.70

 unsolved

-0.61

 rites

-0.61

 underworld

-0.59

utch

-0.59

 gulf

-0.58

 rehab

-0.58

 dedication

-0.57

FU

-0.56

uilt

-0.56

POSITIVE LOGITS

Vox

1.29

 Machina

1.08

 Emin

0.93

 Pengu

0.89

 popul

0.83

odox

0.83

gil

0.81

NV

0.77

 Nash

0.77

Care

0.73

Activations Density 0.009%

proper nouns, particularly from the same source

mentions of the media outlet "Vox"

No Comments

No Known Activations