INDEX

Explanations

words related to destruction or harm

oai_token-act-pair · gpt-3.5-turbo

various word stems or roots

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

 trump

-0.64

Byr

-0.61

shire

-0.61

 Creed

-0.61

hz

-0.61

 reckoning

-0.58

 gasp

-0.56

awed

-0.55

 Guinness

-0.55

 Duke

-0.55

POSITIVE LOGITS

ļéĨĴ

0.93

ĸ

0.83

eties

0.79

portation

0.77

Ħ

0.76

hett

0.74

etary

0.74

apers

0.72

į

0.69

essa

0.69

Activations Density 0.273%

words related to destruction or harm

various word stems or roots

No Comments

No Known Activations