INDEX

Explanations

phrases related to incidents or events with negative outcomes, such as accidents, deaths, and damage

oai_token-act-pair · gpt-3.5-turbo

phrases that describe events resulting in injuries or fatalities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iths

-0.80

Same

-0.76

atech

-0.74

ith

-0.74

 nurture

-0.72

beh

-0.71

wcsstore

-0.71

 behaves

-0.68

acy

-0.66

thren

-0.65

POSITIVE LOGITS

 deaths

1.14

 fatalities

1.13

 amput

1.06

 death

0.96

 hospitalized

0.94

 injuring

0.93

ãĥĩãĤ£

0.90

 injuries

0.89

 casualties

0.88

 miscarriage

0.84

Activations Density 0.272%

phrases related to incidents or events with negative outcomes, such as accidents, deaths, and damage

phrases that describe events resulting in injuries or fatalities

No Comments

No Known Activations