INDEX

Explanations

terms related to negative experiences or actions, especially involving lack of consent

oai_token-act-pair · gpt-3.5-turbo

terms related to unwanted situations or outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

alach

-0.91

ophers

-0.83

bleacher

-0.80

sis

-0.80

inet

-0.79

etics

-0.79

imore

-0.77

yi

-0.76

ership

-0.76

olitan

-0.76

POSITIVE LOGITS

 pregnancies

1.15

 unwanted

0.98

 pregnancy

0.94

 consequences

0.87

 Parenthood

0.86

ãĤī

0.85

 consequence

0.84

 adolesc

0.82

 unintended

0.81

 baggage

0.78

Activations Density 0.018%

terms related to negative experiences or actions, especially involving lack of consent

terms related to unwanted situations or outcomes

No Comments

No Known Activations