INDEX

Explanations

profane and aggressive language

oai_token-act-pair · gpt-3.5-turbo

profanity and derogatory terms

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

 striking

-0.77

 Annotations

-0.71

 Canaver

-0.71

 comprehens

-0.67

 Emanuel

-0.67

 Ambro

-0.67

 prolonged

-0.66

 departures

-0.65

 descending

-0.64

 statically

-0.64

POSITIVE LOGITS

****

1.25

ookie

1.23

*****

1.18

aaaa

1.18

itty

1.15

anky

1.11

oooo

1.11

aaa

1.10

fuck

1.08

$$

1.08

Activations Density 0.279%

profane and aggressive language

profanity and derogatory terms

No Comments

No Known Activations