INDEX

Explanations

the word 'Ļ'

oai_token-act-pair · gpt-3.5-turbo

expressions indicating a sense of obligation or necessity

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

 Drawn

-0.81

Yen

-0.69

 Antar

-0.67

 Downs

-0.67

 tagged

-0.63

ADS

-0.62

 Races

-0.62

 partners

-0.61

 Berk

-0.61

 Mobil

-0.61

POSITIVE LOGITS

doesn

1.11

little

1.10

problem

1.09

thing

1.07

might

1.05

could

1.05

mma

1.04

would

1.03

something

1.03

makes

1.02

Activations Density 0.097%

the word 'Ļ'

expressions indicating a sense of obligation or necessity

No Comments

No Known Activations