INDEX

Explanations

terms related to algorithmic sets, such as Mandelbrot sets

oai_token-act-pair · gpt-3.5-turbo

the presence of a specific proper noun

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 destro

-0.83

£ı

-0.81

 tremend

-0.81

 behavi

-0.77

ĪĴ

-0.75

 exha

-0.74

 resil

-0.74

 mosqu

-0.71

 suspic

-0.70

inventoryQuantity

-0.70

POSITIVE LOGITS

ronics

1.12

ractor

1.11

ruck

1.09

ribute

1.05

olkien

1.05

ract

1.03

iful

1.03

ween

1.03

oxin

1.00

uple

0.99

Activations Density 0.068%

terms related to algorithmic sets, such as Mandelbrot sets

the presence of a specific proper noun

No Comments

No Known Activations