INDEX

Explanations

phrases related to being correct or accurate

oai_token-act-pair · gpt-3.5-turbo

phrases related to correctness or achieving the right outcome

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ĸļ

-0.94

hyde

-0.77

ility

-0.70

archive

-0.64

ulz

-0.62

EEK

-0.62

mop

-0.60

 Chic

-0.60

 Sawyer

-0.59

 whine

-0.59

POSITIVE LOGITS

eous

1.26

fully

0.84

headed

0.83

nered

0.75

footed

0.74

ened

0.73

sky

0.72

este

0.71

 wing

0.71

ness

0.70

Activations Density 0.047%

phrases related to being correct or accurate

phrases related to correctness or achieving the right outcome

No Comments

No Known Activations