INDEX

Explanations

the concept of something being the "first" in a sequence or series

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the concept of "first."

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

tics

-0.83

gery

-0.70

mund

-0.67

morph

-0.66

sav

-0.64

Ïī

-0.64

ingen

-0.62

bos

-0.61

inside

-0.61

 Mahm

-0.61

POSITIVE LOGITS

 baseman

1.05

 responders

1.04

 batch

0.91

 installment

0.89

 lady

0.86

 tentative

0.83

 foray

0.81

 glimpse

0.81

 wave

0.80

 impressions

0.78

Activations Density 0.089%

the concept of something being the "first" in a sequence or series

phrases indicating the concept of "first."

No Comments

No Known Activations