INDEX

Explanations

contractions of the form "can't"

oai_token-act-pair · gpt-3.5-turbo

phrases expressing disbelief or strong negation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

imity

-0.66

senal

-0.64

 Appears

-0.64

CI

-0.63

ItemImage

-0.63

 bount

-0.62

estate

-0.61

amer

-0.61

grounds

-0.60

 Reborn

-0.59

POSITIVE LOGITS

 afford

1.36

 imagine

1.01

 cope

0.92

 conceive

0.91

 reproduce

0.90

 tolerate

0.90

 muster

0.89

 handle

0.87

get

0.86

 feas

0.86

Activations Density 0.044%

contractions of the form "can't"

phrases expressing disbelief or strong negation

No Comments

No Known Activations