INDEX

Explanations

references to power capacity measurements, particularly in gigawatts

oai_token-act-pair · gpt-3.5-turbo

references to gigawatts and related energy metrics

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 tender

-0.76

motion

-0.73

 salt

-0.68

 immobil

-0.67

 motion

-0.66

 Milk

-0.65

pper

-0.65

 Motion

-0.64

 Dairy

-0.64

 Polar

-0.63

POSITIVE LOGITS

GW

3.94

GW

3.89

MW

1.33

GG

1.27

BG

1.21

MW

1.13

WB

1.11

awatts

1.05

WB

1.03

GBT

1.03

Activations Density 0.014%

references to power capacity measurements, particularly in gigawatts

references to gigawatts and related energy metrics

No Comments

No Known Activations