INDEX

Explanations

the word "Ir" with a high activation value

oai_token-act-pair · gpt-3.5-turbo

mentions of "Ir" (likely a reference to a specific character or entity) throughout the document

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

pop

-0.73

 welcome

-0.70

pop

-0.70

 solo

-0.70

ality

-0.65

 chambers

-0.65

 popping

-0.65

ango

-0.64

MON

-0.64

alog

-0.63

POSITIVE LOGITS

Ir

3.70

Ir

2.56

ir

1.44

Il

1.30

Er

1.30

IR

1.28

 Irving

1.28

Kir

1.16

 Imper

1.15

 Irwin

1.15

Activations Density 0.017%

the word "Ir" with a high activation value

mentions of "Ir" (likely a reference to a specific character or entity) throughout the document

No Comments

No Known Activations