INDEX

Explanations

words related to planning or scheming

oai_token-act-pair · gpt-3.5-turbo

the term "cons," indicating discussions about conspiracies or collusion

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 calves

-0.64

 bugs

-0.63

 Waters

-0.62

 Meadow

-0.59

 stakes

-0.59

 Welsh

-0.58

 Pyth

-0.58

lihood

-0.57

ECO

-0.57

 macros

-0.57

POSITIVE LOGITS

cription

1.35

cript

1.33

oling

1.27

ignment

1.27

pire

1.27

oled

1.24

pired

1.22

igl

1.18

igned

1.16

uls

1.16

Activations Density 0.013%

words related to planning or scheming

the term "cons," indicating discussions about conspiracies or collusion

No Comments

No Known Activations