INDEX

Explanations

names of people or places

oai_token-act-pair · gpt-3.5-turbo

specific letters or character sequences that could be part of proper nouns or titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Snap

-0.75

Zap

-0.73

PL

-0.73

 Ples

-0.72

pl

-0.71

 therap

-0.70

Cho

-0.69

bol

-0.68

wra

-0.67

 Planned

-0.66

POSITIVE LOGITS

ient

1.35

ien

1.31

ison

1.15

ioxide

1.14

ius

1.14

ieu

1.12

iot

1.11

ios

1.11

ion

1.09

iber

1.09

Activations Density 0.205%

names of people or places

specific letters or character sequences that could be part of proper nouns or titles

No Comments

No Known Activations