INDEX

Explanations

phrases related to casualties or damages, especially in a negative context

oai_token-act-pair · gpt-3.5-turbo

terms related to casualties and related incidents

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Els

-0.76

 tokens

-0.69

 guests

-0.68

Ku

-0.67

 signatures

-0.66

 wizards

-0.65

masters

-0.64

Eld

-0.64

 Legion

-0.64

 queens

-0.64

POSITIVE LOGITS

 casualty

3.33

 Foley

1.84

arella

1.52

oley

1.06

 paycheck

1.05

 footnote

0.99

 vacancy

0.90

 vict

0.87

 Casual

0.87

 battleground

0.86

Activations Density 0.040%

phrases related to casualties or damages, especially in a negative context

terms related to casualties and related incidents

No Comments

No Known Activations