INDEX

Explanations

words related to betrayal and dishonesty

oai_token-act-pair · gpt-3.5-turbo

the prefix "un" indicating negation or reversal of a quality or state

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

uyomi

-1.03

anwhile

-0.94

å§«

-0.89

hyde

-0.88

Downloadha

-0.76

 Nanto

-0.75

 Nightmares

-0.68

 Opera

-0.67

*/(

-0.67

 Thornton

-0.67

POSITIVE LOGITS

avour

1.11

athom

1.10

riend

1.00

ishable

0.98

ortun

0.96

requ

0.96

ortunately

0.96

ruit

0.95

ixed

0.95

itt

0.89

Activations Density 0.003%

words related to betrayal and dishonesty

the prefix "un" indicating negation or reversal of a quality or state

No Comments

No Known Activations