INDEX

Explanations

phrases related to being taken by surprise or experiencing unexpected events

oai_token-act-pair · gpt-3.5-turbo

phrases that express emotional reactions or sentiments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

•layer 10 - surprise

No Comments

Negative Logits

 Slate

-0.85

 Whedon

-0.84

 BuzzFeed

-0.78

 Isles

-0.77

 Holt

-0.76

 Moff

-0.76

 Hyde

-0.74

 Scott

-0.73

 Doyle

-0.72

 Huff

-0.72

POSITIVE LOGITS

 learnt

1.48

 realised

1.12

 envis

1.05

 till

1.03

Â´

1.02

 realise

1.02

 analysed

1.00

alore

0.99

 organise

0.99

jriwal

0.98

Activations Density 0.817%

phrases related to being taken by surprise or experiencing unexpected events

phrases that express emotional reactions or sentiments

No Comments

No Known Activations