INDEX

Explanations

the word "bro" with varying contexts

oai_token-act-pair · gpt-3.5-turbo

repeated informal address terms, specifically "bro."

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

ById

-0.70

Fra

-0.69

":["

-0.69

ARB

-0.68

Catalog

-0.68

INAL

-0.67

DEP

-0.67

LP

-0.66

LI

-0.66

ENC

-0.66

POSITIVE LOGITS

swer

1.09

chet

0.94

thel

0.90

idered

0.89

ellery

0.83

oding

0.83

bro

0.83

roach

0.82

holes

0.81

oks

0.81

Activations Density 0.012%

the word "bro" with varying contexts

repeated informal address terms, specifically "bro."

No Comments

No Known Activations