INDEX

Explanations

text related to changes or modifications, specifically words prefixed with "re"

oai_token-act-pair · gpt-3.5-turbo

instances of the prefix "re" suggesting a focus on actions or concepts related to repetition, restoration, or renewal

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

BuyableInstoreAndOnline

-0.93

DragonMagazine

-0.82

 Pebble

-0.76

 Beir

-0.75

 Kafka

-0.71

 Nest

-0.68

gag

-0.66

 Zamb

-0.66

 Phar

-0.64

 Philips

-0.63

POSITIVE LOGITS

venge

1.14

becca

1.11

actor

1.08

aper

1.07

ached

1.07

apers

1.05

porter

1.05

venue

1.01

versible

1.00

pos

0.98

Activations Density 0.043%

text related to changes or modifications, specifically words prefixed with "re"

instances of the prefix "re" suggesting a focus on actions or concepts related to repetition, restoration, or renewal

No Comments

No Known Activations