INDEX

Explanations

the word "pending" followed by a number

oai_token-act-pair · gpt-3.5-turbo

references to ongoing legal or judicial processes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

anners

-0.73

iuses

-0.63

 Actually

-0.62

 Advent

-0.61

guide

-0.60

istics

-0.59

tymology

-0.58

 Literally

-0.58

hyper

-0.57

 Sailor

-0.57

POSITIVE LOGITS

 pending

3.84

 awaiting

1.66

 unresolved

1.36

 impending

1.35

 ongoing

1.26

 awaited

1.24

 forthcoming

1.23

 unsolved

1.18

 looming

1.17

 imminent

1.14

Activations Density 0.010%

the word "pending" followed by a number

references to ongoing legal or judicial processes

No Comments

No Known Activations