INDEX

Explanations

phrases related to negative or unwanted situations or conditions

oai_token-act-pair · gpt-3.5-turbo

references to unwanted situations or actions

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

bleacher

-0.91

bard

-0.88

alach

-0.86

hetti

-0.85

ramid

-0.84

iverpool

-0.80

olitan

-0.79

mberg

-0.78

ophers

-0.78

opsis

-0.77

POSITIVE LOGITS

 pregnancies

1.19

 Parenthood

0.99

 pregnancy

0.98

 unwanted

0.96

ãĤī

0.80

nu

0.75

 baggage

0.74

 adolesc

0.72

 burdens

0.71

 unintended

0.71

Activations Density 0.028%

phrases related to negative or unwanted situations or conditions

references to unwanted situations or actions

No Comments

No Known Activations