INDEX

Explanations

phrases related to very small details or nuances

oai_token-act-pair · gpt-3.5-turbo

phrases related to minimal thresholds or subtle details

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

anwhile

-0.78

mbuds

-0.77

axter

-0.74

 Journals

-0.68

olan

-0.66

archives

-0.65

 Centers

-0.64

building

-0.63

aver

-0.63

ingen

-0.62

POSITIVE LOGITS

 slightest

1.23

 Subtle

0.90

iculty

0.89

 hint

0.86

 whiff

0.82

terness

0.79

 imaginable

0.79

 conceivable

0.78

00200000

0.77

elligent

0.75

Activations Density 0.007%

phrases related to very small details or nuances

phrases related to minimal thresholds or subtle details

No Comments

No Known Activations