INDEX

Explanations

phrases describing methods or processes

oai_token-act-pair · gpt-3.5-turbo

instances of the word "by" and related phrases indicating processes or methods

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

gow

-0.86

ppa

-0.83

qqa

-0.77

ijah

-0.74

reb

-0.74

itled

-0.73

roller

-0.71

stead

-0.71

eah

-0.69

ahead

-0.68

POSITIVE LOGITS

 virtue

1.22

products

1.02

 multiplying

0.99

 fiat

0.84

laws

0.79

catch

0.77

 removing

0.76

 default

0.75

 applying

0.75

gone

0.74

Activations Density 0.166%

phrases describing methods or processes

instances of the word "by" and related phrases indicating processes or methods

No Known Activations