INDEX

Explanations

adjectives or phrases related to power or strength

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

edIn

-0.74

 Seym

-0.72

gow

-0.68

assetsadobe

-0.64

 Greenwald

-0.63

 Lauder

-0.61

 unborn

-0.61

OPLE

-0.60

Reloaded

-0.60

 Retrieved

-0.60

POSITIVE LOGITS

imposed

1.33

nova

1.31

visor

1.29

visory

1.20

charg

1.16

visors

1.14

charged

1.12

intend

1.09

cedes

1.07

cell

1.07

Activations Density 0.396%

adjectives or phrases related to power or strength

No Comments

No Known Activations