INDEX

Explanations

phrases contrasting two different concepts or ideas

oai_token-act-pair · gpt-3.5-turbo

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pert

-0.69

zero

-0.68

meta

-0.66

und

-0.64

SI

-0.62

estic

-0.61

ILLE

-0.61

wake

-0.61

emn

-0.60

est

-0.60

POSITIVE LOGITS

 rather

1.94

rather

1.50

 instead

1.49

 Rather

1.37

instead

1.18

Rather

1.16

 Instead

1.14

Instead

1.13

 nevertheless

1.13

 nonetheless

1.09

Activations Density 0.097%

phrases contrasting two different concepts or ideas

No Comments

No Known Activations