INDEX

Explanations

the word "didn't" followed by personal reflections or statements

oai_token-act-pair · gpt-3.5-turbo

instances of negation or expressions showing what the subject did not do

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 populated

-0.72

 retirees

-0.68

 couch

-0.66

 protected

-0.65

 shelves

-0.62

 mutually

-0.62

 adversaries

-0.61

 disabled

-0.61

 Anarchy

-0.60

 retired

-0.58

POSITIVE LOGITS

't

1.58

ÃŃ

1.05

Â´

0.96

0.94

ned

0.91

nt

0.89

gered

0.88

iting

0.88

uts

0.88

ovan

0.86

Activations Density 0.083%

the word "didn't" followed by personal reflections or statements

instances of negation or expressions showing what the subject did not do

No Comments

No Known Activations