INDEX

Explanations

information about crime or accidents, focusing on acts like being shot, being set on fire, and police investigations

oai_token-act-pair · gpt-3.5-turbo

events related to violent incidents or crimes

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

limits

-0.81

notes

-0.77

æ©Ł

-0.70

izons

-0.70

attr

-0.69

paren

-0.69

Critics

-0.69

imon

-0.67

yright

-0.67

Tokens

-0.66

POSITIVE LOGITS

 intoxicated

0.98

 explosives

0.98

 homicide

0.97

 burg

0.96

 motive

0.93

 Paddock

0.93

 abducted

0.92

 foul

0.91

 drunk

0.88

 detonated

0.88

Activations Density 0.364%

information about crime or accidents, focusing on acts like being shot, being set on fire, and police investigations

events related to violent incidents or crimes

No Comments

No Known Activations