INDEX

Explanations

the word "ito" followed by various different prefixes

oai_token-act-pair · gpt-3.5-turbo

repeated references to specific names or terms ending with "ito"

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

 distance

-0.70

 dawn

-0.64

 ribbon

-0.64

 above

-0.63

 divers

-0.63

 foot

-0.63

 resolution

-0.61

 gate

-0.60

 Lyme

-0.59

 bees

-0.59

POSITIVE LOGITS

ito

4.59

ita

1.64

itus

1.42

itas

1.31

ido

1.26

iko

1.17

ato

1.14

ico

1.14

uto

1.12

illo

1.12

Activations Density 0.016%

the word "ito" followed by various different prefixes

repeated references to specific names or terms ending with "ito"

No Comments

No Known Activations