INDEX

Explanations

phrases indicating personal opinions or beliefs

oai_token-act-pair · gpt-3.5-turbo

questions directed at an audience or expressions of uncertainty regarding opinions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

alore

-0.67

 Guests

-0.63

oga

-0.60

 Wonderful

-0.58

onday

-0.56

dyl

-0.54

 guests

-0.53

photos

-0.53

aceae

-0.53

dimension

-0.53

POSITIVE LOGITS

isSpecialOrderable

0.60

ctory

0.56

rx

0.55

ï¸ı

0.55

retty

0.53

reads

0.53

plin

0.53

ription

0.51

metic

0.51

vu

0.50

Activations Density 0.193%

phrases indicating personal opinions or beliefs

questions directed at an audience or expressions of uncertainty regarding opinions

No Comments

No Known Activations