INDEX

Explanations

words that indicate things are related, connected, or similar

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "sameness" or being 'the same'

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

 Provided

-0.69

phasis

-0.68

live

-0.66

reetings

-0.63

Bei

-0.63

skirts

-0.63

aze

-0.62

orsi

-0.61

better

-0.61

members

-0.59

POSITIVE LOGITS

 vein

1.24

 exact

0.97

 thing

0.95

 timeframe

0.92

 amount

0.86

 category

0.82

 kind

0.82

 manner

0.79

 scenario

0.77

 ballpark

0.77

Activations Density 0.046%

words that indicate things are related, connected, or similar

references to the concept of "sameness" or being 'the same'

No Comments

No Known Activations