INDEX

Explanations

words related to genuineness or compliance with truth

oai_token-act-pair · gpt-3.5-turbo

references to authenticity and genuine qualities

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

âĢ¢âĢ¢âĢ¢âĢ¢

-0.83

---------

-0.73

EVA

-0.73

 DRAG

-0.72

LAN

-0.71

Various

-0.68

interrupted

-0.68

Pages

-0.65

--+

-0.65

pmwiki

-0.63

POSITIVE LOGITS

 authentic

1.50

 authenticity

1.23

authent

1.18

 authent

1.13

 Authent

1.05

Authent

0.90

atable

0.89

estate

0.82

 recre

0.78

icator

0.78

Activations Density 0.009%

words related to genuineness or compliance with truth

references to authenticity and genuine qualities

No Comments

No Known Activations