INDEX

Explanations

phrases related to violent actions involving the firing of weapons

oai_token-act-pair · gpt-3.5-turbo

instances of the word "open" in various forms related to actions

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

Bey

-0.80

 constitu

-0.74

CTV

-0.73

orum

-0.67

ensis

-0.64

rior

-0.63

cia

-0.63

rogens

-0.61

verages

-0.60

iable

-0.59

POSITIVE LOGITS

 doors

1.21

 Doors

1.03

 portals

0.95

ource

0.93

up

0.90

 wounds

0.85

 door

0.85

 curtains

0.85

tain

0.79

 loopholes

0.78

Activations Density 0.049%

phrases related to violent actions involving the firing of weapons

instances of the word "open" in various forms related to actions

No Comments

No Known Activations