INDEX

Explanations

contractions that include the word "won't"

oai_token-act-pair · gpt-3.5-turbo

negation phrases emphasizing inability or lack of something

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

OTOS

-0.72

illin

-0.68

bian

-0.66

eki

-0.66

gypt

-0.61

 cultivating

-0.61

mens

-0.58

enegger

-0.58

 embodiments

-0.58

MEN

-0.58

POSITIVE LOGITS

't

1.23

itive

0.91

now

0.83

stall

0.82

geon

0.78

Dispatch

0.74

ims

0.73

rar

0.72

æ©

0.72

geons

0.71

Activations Density 0.031%

contractions that include the word "won't"

negation phrases emphasizing inability or lack of something

No Comments

No Known Activations