INDEX

Explanations

phrases related to analysis or comparison of different aspects

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate relationships or conditions involving the word "with."

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

hood

-0.74

sat

-0.66

fair

-0.64

fiction

-0.62

deen

-0.61

sov

-0.60

she

-0.60

oult

-0.59

soft

-0.59

berto

-0.59

POSITIVE LOGITS

 regards

1.96

 regard

1.89

 respect

1.56

stood

1.48

draw

1.44

standing

1.25

 impunity

1.22

drawn

1.21

holding

1.15

 hindsight

0.98

Activations Density 0.244%

phrases related to analysis or comparison of different aspects

phrases that indicate relationships or conditions involving the word "with."

No Comments

No Known Activations