INDEX

Explanations

phrases related to pairs or couples

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

schild

-0.75

ADRA

-0.67

inez

-0.62

Lenin

-0.60

orgetown

-0.58

 Depot

-0.56

ESCO

-0.56

CLASS

-0.56

INA

-0.56

 Directorate

-0.55

POSITIVE LOGITS

ings

1.00

lihood

0.89

wise

0.82

hood

0.82

 mates

0.81

footed

0.81

atures

0.78

osaurs

0.77

pair

0.77

mates

0.72

Activations Density 0.052%

phrases related to pairs or couples

No Comments

No Known Activations