INDEX

Explanations

references to brothers

oai_token-act-pair · gpt-3.5-turbo

mentions of siblings, particularly brothers

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

issions

-0.79

Population

-0.76

orney

-0.73

itures

-0.73

USE

-0.71

veyard

-0.69

erest

-0.69

argon

-0.69

alling

-0.67

pmwiki

-0.66

POSITIVE LOGITS

hood

1.41

 brother

0.97

 brothers

0.93

brother

0.90

friend

0.87

heses

0.83

Brother

0.82

pins

0.79

hesis

0.79

 uncle

0.78

Activations Density 0.014%

references to brothers

mentions of siblings, particularly brothers

No Comments

No Known Activations