INDEX

Explanations

words related to tragedies or unfortunate incidents, particularly related to death

oai_token-act-pair · gpt-3.5-turbo

references to death and its related circumstances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

umar

-0.74

Cola

-0.74

 Libre

-0.74

amaru

-0.70

MN

-0.67

icons

-0.67

toe

-0.66

orers

-0.66

EEK

-0.66

DIR

-0.64

POSITIVE LOGITS

bed

1.05

 toll

1.04

 certificate

0.91

 sentence

0.90

blow

0.87

adder

0.86

 penalty

0.86

 threats

0.85

 Penalty

0.83

 certificates

0.83

Activations Density 0.053%

words related to tragedies or unfortunate incidents, particularly related to death

references to death and its related circumstances

No Comments

No Known Activations