INDEX

Explanations

principles (moral, legal, or fundamental truths)

oai_token-act-pair · gpt-3.5-turbo

references to 'principle'

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 tagged

-0.70

jer

-0.68

 outs

-0.65

acked

-0.65

 Crash

-0.64

iro

-0.64

asks

-0.63

irl

-0.63

pour

-0.62

 Bust

-0.62

POSITIVE LOGITS

 principle

3.73

 Principle

2.48

 principles

2.20

 Principles

1.61

 premise

1.50

 theory

1.45

 theorem

1.41

 prin

1.40

 concept

1.36

 principals

1.31

Activations Density 0.005%

principles (moral, legal, or fundamental truths)

references to 'principle'

No Comments

No Known Activations