INDEX

Explanations

words related to ignition sources, such as "spark."

oai_token-act-pair · gpt-3.5-turbo

references to the concept of a "spark" in various contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Orth

-0.81

 Browne

-0.72

 Dietary

-0.71

 orth

-0.70

li

-0.70

ulo

-0.70

atos

-0.69

Ass

-0.68

Board

-0.68

 Corrections

-0.65

POSITIVE LOGITS

 spark

3.86

 sparks

2.24

 Spark

1.88

 ignite

1.59

 sparking

1.53

 catalyst

1.41

 cataly

1.39

 spur

1.31

 Sparks

1.30

 inspire

1.22

Activations Density 0.026%

words related to ignition sources, such as "spark."

references to the concept of a "spark" in various contexts

No Comments

No Known Activations