INDEX

Explanations

offensive and derogatory terms

oai_token-act-pair · gpt-3.5-turbo

terms related to sexuality and derogatory language

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

ulhu

-0.84

actic

-0.81

scl

-0.79

owan

-0.79

sonian

-0.77

son

-0.74

 Flavoring

-0.73

ointment

-0.73

ERG

-0.72

acter

-0.71

POSITIVE LOGITS

 panties

0.94

 nuns

0.80

 pussy

0.78

 lips

0.77

 vagina

0.76

 Melania

0.76

 breasts

0.74

 Lucia

0.73

 boobs

0.73

 Riot

0.72

Activations Density 0.054%

offensive and derogatory terms

terms related to sexuality and derogatory language

No Comments

No Known Activations