INDEX

Explanations

terms related to sub-categories or specific features within a broader category

oai_token-act-pair · gpt-3.5-turbo

references to subdivisions or subcategories of various topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 dearly

-0.71

 flared

-0.69

 Polk

-0.69

dur

-0.65

 Reloaded

-0.64

 firsthand

-0.63

 greeted

-0.62

 dentist

-0.62

 outweigh

-0.61

 Canter

-0.61

POSITIVE LOGITS

Saharan

1.50

surface

1.34

zero

1.32

division

1.32

paragraph

1.31

contract

1.29

committee

1.29

standard

1.27

div

1.26

continental

1.25

Activations Density 0.024%

terms related to sub-categories or specific features within a broader category

references to subdivisions or subcategories of various topics

No Comments

No Known Activations