INDEX

Explanations

references to the word "Sand" followed by a number

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Sand" as a significant keyword

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

urses

-0.89

uates

-0.81

mercial

-0.78

olate

-0.75

UTERS

-0.74

KT

-0.74

 conclud

-0.73

berus

-0.72

Reloaded

-0.71

merce

-0.69

POSITIVE LOGITS

wic

1.02

usky

1.00

alph

0.96

hya

0.90

oval

0.90

paper

0.89

box

0.86

hill

0.85

mallow

0.84

er

0.84

Activations Density 0.014%

references to the word "Sand" followed by a number

occurrences of the word "Sand" as a significant keyword

No Comments

No Known Activations