INDEX

Explanations

words related to assigning blame or identifying a perpetrator, particularly with negative connotations

oai_token-act-pair · gpt-3.5-turbo

references to individuals or entities identified as 'culprits' or 'masterminds.'

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

forms

-0.76

ISTER

-0.74

ORT

-0.73

intend

-0.73

cyclop

-0.73

formed

-0.72

pec

-0.69

porting

-0.69

verend

-0.68

 Parables

-0.68

POSITIVE LOGITS

 culprit

1.21

æ©

0.78

ãĥ¼ãĥĨãĤ£

0.78

 mastermind

0.76

0.74

ãĤ¸

0.70

enary

0.68

 perpetrator

0.68

TextColor

0.66

ãĥĨ

0.66

Activations Density 0.009%

words related to assigning blame or identifying a perpetrator, particularly with negative connotations

references to individuals or entities identified as 'culprits' or 'masterminds.'

No Comments

No Known Activations