INDEX

Explanations

references to different kinds of crimes, particularly those against humanity

oai_token-act-pair · gpt-3.5-turbo

references to crimes against humanity and war crimes

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

pole

-0.79

alg

-0.78

quickShipAvailable

-0.77

zl

-0.69

visors

-0.64

agin

-0.63

ulous

-0.63

ersed

-0.61

 snipp

-0.61

rition

-0.61

POSITIVE LOGITS

 committed

1.09

 perpetrated

0.98

 punishable

0.96

 spree

0.86

 crimes

0.84

 against

0.81

 offences

0.81

 involving

0.80

 unfocusedRange

0.79

 victim

0.75

Activations Density 0.032%

references to different kinds of crimes, particularly those against humanity

references to crimes against humanity and war crimes

No Comments

No Known Activations