INDEX

Explanations

memories

oai_token-act-pair · gpt-3.5-turbo

references to memory or memoir-related content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ASC

-0.76

 Icelandic

-0.70

 Caribbean

-0.70

 Strauss

-0.67

 boiling

-0.66

 Nordic

-0.64

 Ducks

-0.64

 Jewish

-0.64

 Judaism

-0.63

BDS

-0.62

POSITIVE LOGITS

nesota

1.26

mem

1.25

phis

1.08

Mem

1.08

oleon

1.07

eor

1.05

orable

1.04

oir

1.03

oria

1.01

pty

0.97

Activations Density 0.006%

memories

references to memory or memoir-related content

No Comments

No Known Activations