INDEX

Explanations

phrases indicating the ability or potential for action

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.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

ĪĴ

-0.76

quickShipAvailable

-0.70

lights

-0.67

 Explan

-0.66

furt

-0.66

lies

-0.65

rant

-0.64

laughter

-0.61

listed

-0.61

 Closing

-0.61

POSITIVE LOGITS

 communicate

1.32

 manipulate

1.30

 customize

1.24

 achieve

1.21

 differentiate

1.21

 concentrate

1.20

 utilize

1.19

 capitalize

1.16

 cope

1.16

 evade

1.15

Activations Density 0.125%

phrases indicating the ability or potential for action

No Comments

No Known Activations