INDEX

Explanations

actionable phrases or directives

oai_token-act-pair · gpt-3.5-turbo

imperative phrases and commands

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

ersen

-0.51

inous

-0.50

bys

-0.49

 plent

-0.48

pez

-0.48

odox

-0.48

eatured

-0.47

izoph

-0.47

uesday

-0.46

eatures

-0.45

POSITIVE LOGITS

to

0.82

To

0.73

to

0.72

To

0.71

.*

0.65

.''.

0.65

Created

0.62

.''

0.61

0.60

Activations Density 0.984%

actionable phrases or directives

imperative phrases and commands

No Comments

No Known Activations