INDEX

Explanations

the phrase "going to."

oai_token-act-pair · gpt-3.5-turbo

negative statements about intention or action

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

origin

-0.69

 widest

-0.67

ortment

-0.65

çīĪ

-0.63

DragonMagazine

-0.63

LOS

-0.62

EB

-0.61

nen

-0.60

iang

-0.59

å§«

-0.59

POSITIVE LOGITS

 anymore

1.29

 necessarily

1.02

 whatsoever

0.97

nor

0.96

 slightest

0.93

 anything

0.91

 bother

0.89

 anybody

0.88

 anywhere

0.87

 bothered

0.85

Activations Density 0.101%

the phrase "going to."

negative statements about intention or action

No Comments

No Known Activations