INDEX

Explanations

instances of the word "no" followed by a number representing the strength of the absence indicated

oai_token-act-pair · gpt-3.5-turbo

instances of the word "no" indicating absence or lack

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 unilaterally

-0.78

 jointly

-0.75

ARY

-0.65

Bloomberg

-0.64

completely

-0.64

 Exclusive

-0.64

 unanimously

-0.63

 Historically

-0.63

intensive

-0.62

Ĥİ

-0.62

POSITIVE LOGITS

oses

1.13

 clue

1.04

xious

1.01

 idea

0.90

 trace

0.90

 explanation

0.89

 indication

0.88

isy

0.84

 recollection

0.84

 luck

0.83

Activations Density 0.087%

instances of the word "no" followed by a number representing the strength of the absence indicated

instances of the word "no" indicating absence or lack

No Comments

No Known Activations