INDEX

Explanations

numbers

oai_token-act-pair · gpt-3.5-turbo

references to familial terms or relationships

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

hedral

-0.71

rage

-0.66

stab

-0.64

odcast

-0.62

oint

-0.62

ache

-0.61

 handc

-0.61

PRES

-0.61

 hostage

-0.61

FER

-0.61

POSITIVE LOGITS

çļ

0.74

ãĥĨãĤ£

0.73

UI

0.69

ãĥĩãĤ£

0.69

 Wonders

0.69

estones

0.67

vale

0.67

ãĤ¬

0.66

igans

0.66

 Streets

0.65

Activations Density 0.000%

numbers

references to familial terms or relationships

No Comments

No Known Activations