INDEX

Explanations

two consecutive capital letters

oai_token-act-pair · gpt-3.5-turbo

headings or sections in a structured document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 eleph

-0.83

 crunch

-0.79

 enriched

-0.79

 autom

-0.78

 neighb

-0.78

 speeding

-0.77

 cement

-0.77

 thumbs

-0.76

 grip

-0.76

 freezer

-0.75

POSITIVE LOGITS

Main

1.47

General

1.46

Various

1.44

There

1.42

Information

1.41

This

1.41

None

1.40

Early

1.39

The

1.39

See

1.38

Activations Density 0.143%

two consecutive capital letters

headings or sections in a structured document

No Comments

No Known Activations