INDEX

Explanations

words ending in 'ed'

oai_token-act-pair · gpt-3.5-turbo

the occurrence of the abbreviation "ED" used in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 persu

-0.69

 patron

-0.68

lis

-0.65

Gon

-0.62

 Donn

-0.60

hei

-0.59

wives

-0.59

 Rath

-0.57

 worth

-0.56

 Wein

-0.56

POSITIVE LOGITS

ED

3.97

ed

1.87

ING

1.82

eds

1.69

ES

1.63

EMENT

1.61

IED

1.58

ABLE

1.55

ERS

1.53

EST

1.51

Activations Density 0.015%

words ending in 'ed'

the occurrence of the abbreviation "ED" used in various contexts

No Comments

No Known Activations