INDEX

Explanations

phrases indicating the beginning or starting of something

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the initiation or progression of events or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 oversaw

-0.70

 Palest

-0.65

CI

-0.65

 avoided

-0.64

done

-0.64

 cares

-0.63

keeping

-0.62

 supervised

-0.61

 congratulated

-0.60

didn

-0.60

POSITIVE LOGITS

 crumble

1.30

 emerge

1.24

 unfold

1.22

 fade

1.20

 explode

1.13

 dawn

1.12

 trickle

1.11

 sink

1.11

 dissolve

1.08

 fray

1.08

Activations Density 0.162%

phrases indicating the beginning or starting of something

phrases indicating the initiation or progression of events or conditions

No Comments

No Known Activations