INDEX

Explanations

technical terms or product names

oai_token-act-pair · gpt-3.5-turbo

references to "Pro" or professional terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

assetsadobe

-0.78

ãĤ¼ãĤ¦ãĤ¹

-0.76

ĸļ

-0.76

bodied

-0.73

gow

-0.72

 fault

-0.69

 needles

-0.69

worms

-0.66

 weave

-0.65

 Norn

-0.64

POSITIVE LOGITS

digy

1.24

secution

1.04

perties

1.02

blems

1.02

gressive

0.99

fit

0.98

ceed

0.94

bable

0.91

secut

0.90

posal

0.90

Activations Density 0.007%

technical terms or product names

references to "Pro" or professional terms

No Comments

No Known Activations