INDEX

Explanations

comparisons of things as opposites

oai_token-act-pair · gpt-3.5-turbo

academic or philosophical terms related to contradiction or opposing views

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 corrid

-0.75

 mathemat

-0.65

 Nationwide

-0.65

 stocking

-0.65

 beetles

-0.64

cow

-0.64

 killer

-0.63

 gelatin

-0.62

 congest

-0.61

 sleeves

-0.61

POSITIVE LOGITS

xual

1.41

hesis

1.02

OPLE

1.00

heses

0.96

ynthesis

0.90

hran

0.89

chnology

0.88

ract

0.86

ctr

0.83

olkien

0.83

Activations Density 0.050%

comparisons of things as opposites

academic or philosophical terms related to contradiction or opposing views

No Comments

No Known Activations