INDEX

Explanations

actions or events involving violence or conflict

oai_token-act-pair · gpt-3.5-turbo

phrases related to crises or emergencies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Cosponsors

-0.80

alde

-0.74

iaries

-0.73

sbm

-0.71

uploads

-0.68

raft

-0.67

umerous

-0.66

 redes

-0.65

hent

-0.64

"]=>

-0.63

POSITIVE LOGITS

 somebody

1.49

 someone

1.41

 something

1.21

 your

1.14

someone

1.10

 grandma

1.06

 oneself

1.04

 yourself

1.03

Someone

1.02

 somewhere

1.00

Activations Density 0.668%

actions or events involving violence or conflict

phrases related to crises or emergencies

No Comments

No Known Activations