INDEX

Explanations

mentions of positive attributes or compliments, particularly focusing on the word "cool."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "cool" and its variations in context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 UNITED

-0.73

Reloaded

-0.71

 Examination

-0.69

ãĥĺãĥ©

-0.69

riage

-0.68

Reviewed

-0.67

 Canaver

-0.67

ele

-0.67

Repe

-0.66

uilding

-0.65

POSITIVE LOGITS

 cool

0.91

oola

0.89

ness

0.88

 stuff

0.80

¥µ

0.80

 darn

0.76

stuff

0.76

 gadgets

0.75

 dude

0.75

lahoma

0.74

Activations Density 0.012%

mentions of positive attributes or compliments, particularly focusing on the word "cool."

instances of the word "cool" and its variations in context

No Comments

No Known Activations