INDEX

Explanations

personal pronouns and verbs indicating action

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

-0.74

IOR

-0.70

 departing

-0.65

 Option

-0.65

 Instead

-0.64

lihood

-0.64

uates

-0.63

Instead

-0.62

wik

-0.61

efficients

-0.60

POSITIVE LOGITS

've

1.00

're

0.99

ain

0.86

'll

0.82

 know

0.81

 swear

0.80

 stink

0.76

'm

0.75

 drank

0.75

don

0.74

Activations Density 0.225%

personal pronouns and verbs indicating action

No Comments

No Known Activations