INDEX

Explanations

words related to usefulness or utility

oai_token-act-pair · gpt-3.5-turbo

instances of the word "useful" and related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

uph

-0.68

agate

-0.67

buck

-0.67

uclear

-0.62

BU

-0.62

 Meadows

-0.59

Bam

-0.59

ogenesis

-0.59

á

-0.59

opers

-0.59

POSITIVE LOGITS

 idiots

1.09

 adjunct

0.85

 insights

0.83

 tools

0.80

fully

0.80

tips

0.78

 inform

0.78

tip

0.77

iences

0.76

 insight

0.75

Activations Density 0.057%

words related to usefulness or utility

instances of the word "useful" and related concepts

No Comments

No Known Activations