INDEX

Explanations

phrases indicating control or authority

oai_token-act-pair · gpt-3.5-turbo

phrases that express possession or belonging

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ials

-0.76

riers

-0.68

ities

-0.67

uctions

-0.67

abilities

-0.65

azines

-0.64

itars

-0.63

hran

-0.63

 reperto

-0.61

rollers

-0.61

POSITIVE LOGITS

 sorts

1.21

 course

0.88

 theirs

0.85

course

0.84

icial

0.81

 hers

0.78

 ours

0.74

 mine

0.72

 yours

0.66

rontal

0.65

Activations Density 0.321%

phrases indicating control or authority

phrases that express possession or belonging

No Comments

No Known Activations