INDEX

Explanations

first-person pronouns followed by actions or thoughts

oai_token-act-pair · gpt-3.5-turbo

references to personal pronouns and expressions of self

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 depreciation

-0.64

 ubiqu

-0.64

ahime

-0.63

 diam

-0.62

eday

-0.61

 simultane

-0.60

 skew

-0.59

 enactment

-0.58

artisan

-0.58

 unparalleled

-0.58

POSITIVE LOGITS

soever

0.85

 fuss

0.79

igenous

0.71

OWS

0.68

 cared

0.68

talk

0.67

FO

0.67

rier

0.66

 needed

0.66

verage

0.64

Activations Density 0.049%

first-person pronouns followed by actions or thoughts

references to personal pronouns and expressions of self

No Comments

No Known Activations