INDEX

Explanations

expressions related to negation

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing the presence or state of being of people or things

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

lance

-0.71

 Blades

-0.71

 Locations

-0.64

alities

-0.63

 sorts

-0.61

URN

-0.59

yip

-0.58

 DRAGON

-0.58

 Stras

-0.57

inker

-0.56

POSITIVE LOGITS

 ever

1.50

 EVER

1.39

 Ever

0.95

 harmed

0.94

 necessarily

0.93

 remotely

0.92

 spared

0.89

Ever

0.85

 immune

0.84

 imaginable

0.82

Activations Density 0.077%

expressions related to negation

phrases emphasizing the presence or state of being of people or things

No Comments

No Known Activations