INDEX

Explanations

phrases related to cost or rate statistics

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate a rate or measurement represented in "per" format

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

enegger

-0.74

ARDIS

-0.70

XT

-0.68

 GOODMAN

-0.68

ITED

-0.67

arium

-0.66

Ruk

-0.65

 Unic

-0.64

 Wilhelm

-0.63

ealous

-0.63

POSITIVE LOGITS

 capita

1.41

tains

1.07

ishable

1.06

malink

0.98

manent

0.91

missible

0.88

severe

0.88

missive

0.88

imeter

0.86

uvian

0.84

Activations Density 0.031%

phrases related to cost or rate statistics

phrases that indicate a rate or measurement represented in "per" format

No Comments

No Known Activations