INDEX

Explanations

mentions of self-harm or suicide

oai_token-act-pair · gpt-3.5-turbo

words or terms related to 'sp' that indicate a quality or characteristic

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

DragonMagazine

-0.86

obyl

-0.75

hof

-0.73

 approve

-0.71

 Halls

-0.69

*/(

-0.69

hyde

-0.69

heit

-0.67

 Incarnation

-0.67

 Apprentice

-0.66

POSITIVE LOGITS

atter

1.05

ooky

1.04

atial

1.03

anned

1.00

itt

1.00

amm

0.99

rawl

0.99

rit

0.96

attering

0.95

acious

0.95

Activations Density 0.015%

mentions of self-harm or suicide

words or terms related to 'sp' that indicate a quality or characteristic

No Comments

No Known Activations