INDEX

Explanations

terms related to rankings or positions, particularly with a negative connotation towards being second

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "second largest" or "second most," indicating a comparative ranking

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

pload

-0.65

Ń·

-0.64

 Richards

-0.64

estern

-0.62

aiman

-0.59

 Arabian

-0.59

 Restrict

-0.58

 Cron

-0.58

SOL

-0.58

ÃįÃį

-0.57

POSITIVE LOGITS

degree

1.27

generation

1.23

tier

1.18

chance

1.10

century

1.10

ranked

1.07

level

1.05

dimensional

1.04

rate

1.03

favorite

1.03

Activations Density 0.035%

terms related to rankings or positions, particularly with a negative connotation towards being second

instances of the phrase "second largest" or "second most," indicating a comparative ranking

No Comments

No Known Activations