INDEX

Explanations

names starting with "Jo" in various contexts

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the name "Jo."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

IFIED

-0.77

oÄŁ

-0.72

ashtra

-0.71

 Wikimedia

-0.70

CLASS

-0.70

intendent

-0.68

 amplification

-0.67

enegger

-0.67

flies

-0.66

 overhead

-0.65

POSITIVE LOGITS

Jo

1.31

aquin

1.26

jo

1.17

ining

0.97

Jo

0.96

anna

0.95

zeb

0.93

areth

0.91

ppa

0.91

aqu

0.88

Activations Density 0.009%

names starting with "Jo" in various contexts

repeated mentions of the name "Jo."

No Comments

No Known Activations