INDEX

Explanations

phrases indicating lack of association or connection between different entities or concepts

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing causation or relationships between actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.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

 feasibility

-0.62

riger

-0.60

esthes

-0.59

 holiest

-0.57

ements

-0.56

cember

-0.56

 Gust

-0.54

hooting

-0.53

aniel

-0.53

res

-0.52

POSITIVE LOGITS

 contribute

0.76

 celebrate

0.74

 differentiate

0.73

speak

0.73

 spare

0.73

 prove

0.73

"></

0.72

 satisfy

0.70

 settle

0.70

ensed

0.69

Activations Density 0.061%

phrases indicating lack of association or connection between different entities or concepts

phrases emphasizing causation or relationships between actions

No Comments

No Known Activations