INDEX

Explanations

non-committal phrases following an affirmation

oai_token-act-pair · gpt-3.5-turbo

negative prefixes indicating unfairness or unacceptability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Madden

-0.74

 halves

-0.70

Duo

-0.70

 briefs

-0.67

 slate

-0.65

tee

-0.64

 initials

-0.64

mos

-0.64

 raids

-0.63

 sheet

-0.63

POSITIVE LOGITS

fortunately

1.43

iversity

1.39

usual

1.38

ivers

1.30

stable

1.30

iform

1.29

necessary

1.29

animous

1.27

surprisingly

1.20

employment

1.20

Activations Density 0.021%

non-committal phrases following an affirmation

negative prefixes indicating unfairness or unacceptability

No Comments

No Known Activations