INDEX

Explanations

words related to legal offenses, specifically the term "treason."

oai_token-act-pair · gpt-3.5-turbo

references to treason and betrayal

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Electric

-0.68

gor

-0.67

 Parenthood

-0.67

Roll

-0.66

ĸļ

-0.66

 Gujar

-0.64

arel

-0.63

Pwr

-0.63

 Improvement

-0.63

Ħ

-0.63

POSITIVE LOGITS

ous

1.11

 traitor

1.02

ously

0.97

 treason

0.94

uous

0.86

itors

0.84

 sympath

0.80

warm

0.74

uously

0.73

 guiActiveUn

0.73

Activations Density 0.019%

words related to legal offenses, specifically the term "treason."

references to treason and betrayal

No Comments

No Known Activations