INDEX

Explanations

commands or instructions to do something

oai_token-act-pair · gpt-3.5-turbo

commands or directives that begin with "to."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Appears

-0.85

quickShipAvailable

-0.81

grounds

-0.73

pires

-0.67

majority

-0.64

conom

-0.62

cano

-0.61

ably

-0.61

 threaded

-0.61

yet

-0.61

POSITIVE LOGITS

 evacuate

0.97

 beware

0.91

 avoid

0.90

 obey

0.89

 shove

0.88

 refrain

0.87

 behave

0.86

 conserve

0.86

 protect

0.83

 accompany

0.82

Activations Density 0.098%

commands or instructions to do something

commands or directives that begin with "to."

No Comments

No Known Activations