INDEX

Explanations

names of individuals

oai_token-act-pair · gpt-3.5-turbo

proper names, particularly of people and characters

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

izoph

-0.95

 administ

-0.88

 organis

-0.82

acerb

-0.81

gencies

-0.79

ijing

-0.79

eatures

-0.76

 ascend

-0.73

 util

-0.73

 guiIcon

-0.71

POSITIVE LOGITS

 Watkins

1.21

 Stewart

1.11

 Robinson

1.09

 Williams

1.09

 Roberts

1.07

 Matthews

1.07

 Simmons

1.06

 Boyd

1.04

 Michaels

1.04

 Phillips

1.04

Activations Density 0.333%

names of individuals

proper names, particularly of people and characters

No Comments

No Known Activations