INDEX

Explanations

phrases where someone is speaking or quoting someone else

oai_token-act-pair · gpt-3.5-turbo

quotations or dialogue 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 @ 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

 princip

-0.87

 deterrent

-0.85

adv

-0.79

 frontline

-0.75

 stem

-0.75

 hardened

-0.74

 conve

-0.73

 fluct

-0.73

 vigil

-0.73

 proceeds

-0.72

POSITIVE LOGITS

Hey

1.74

Oh

1.62

hey

1.52

Yeah

1.51

Look

1.47

Fuck

1.44

Okay

1.43

Well

1.42

Damn

1.40

yeah

1.38

Activations Density 0.059%

phrases where someone is speaking or quoting someone else

quotations or dialogue in the text

No Comments

No Known Activations