INDEX

Explanations

instances where products or experiences are described with a focus on their quality or uniqueness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

umably

-0.93

pse

-0.89

 presumably

-0.72

 stranger

-0.70

 gunfire

-0.69

 metaph

-0.69

 epist

-0.69

lier

-0.69

 grop

-0.69

 unlucky

-0.69

POSITIVE LOGITS

 Whether

1.32

 Includes

1.31

 Additionally

1.27

 Designed

1.21

 Simply

1.19

 Featuring

1.19

 Each

1.19

 Besides

1.17

 Learn

1.16

 Available

1.14

Activations Density 0.189%

instances where products or experiences are described with a focus on their quality or uniqueness

No Comments

No Known Activations