INDEX

Explanations

phrases conveying a strong degree of certainty or emphasis, typically using the word "very" or "could very well."

oai_token-act-pair · gpt-3.5-turbo

intensifiers or modifiers, particularly the word "very" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

DragonMagazine

-0.71

oresc

-0.66

BuyableInstoreAndOnline

-0.66

aming

-0.62

 notebooks

-0.60

RAD

-0.59

netflix

-0.58

ONES

-0.56

 Gleaming

-0.56

ative

-0.56

POSITIVE LOGITS

rouse

1.08

venge

1.05

rely

0.96

lege

0.90

arily

0.81

activate

0.80

bably

0.78

own

0.77

fully

0.77

etheless

0.76

Activations Density 0.192%

phrases conveying a strong degree of certainty or emphasis, typically using the word "very" or "could very well."

intensifiers or modifiers, particularly the word "very" and its variations

No Comments

No Known Activations