INDEX

Explanations

references to siblings, especially sisters

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "sister."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

veyard

-0.77

ered

-0.67

upuncture

-0.67

ustomed

-0.65

ocalypse

-0.64

 Lovecraft

-0.63

agn

-0.61

 Yiannopoulos

-0.60

BI

-0.60

atility

-0.60

POSITIVE LOGITS

hood

1.10

 sister

0.98

hips

0.87

heses

0.85

 sisters

0.78

wife

0.76

fax

0.76

hesis

0.76

 brother

0.74

folk

0.74

Activations Density 0.008%

references to siblings, especially sisters

occurrences of the word "sister."

No Comments

No Known Activations