INDEX

Explanations

terms related to file extensions or technical jargon

oai_token-act-pair · gpt-3.5-turbo

abbreviations or symbols indicating expansions or extensions

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

hari

-0.93

Downloadha

-0.78

ker

-0.76

ghai

-0.74

folk

-0.73

maid

-0.72

hart

-0.72

UGH

-0.70

zza

-0.70

gha

-0.68

POSITIVE LOGITS

racted

1.46

ensions

1.46

ension

1.40

ract

1.40

raction

1.29

inct

1.29

ortion

1.22

ractor

1.20

ended

1.18

ractions

1.11

Activations Density 0.011%

terms related to file extensions or technical jargon

abbreviations or symbols indicating expansions or extensions

No Comments

No Known Activations