INDEX

Explanations

phrases or terms within quotations

oai_token-act-pair · gpt-3.5-turbo

quoted phrases or expressions in the text

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

"[

-0.84

 afar

-0.83

 whilst

-0.79

 preceded

-0.78

 Ubisoft

-0.76

 accomp

-0.75

 [âĢ¦]

-0.75

 while

-0.75

 cited

-0.75

 viewed

-0.75

POSITIVE LOGITS

clean

1.46

moral

1.43

death

1.43

safe

1.40

reset

1.39

smart

1.39

zero

1.38

Make

1.37

safety

1.37

comfort

1.36

Activations Density 0.093%

phrases or terms within quotations

quoted phrases or expressions in the text

No Comments

No Known Activations