INDEX

Explanations

phrases indicating reasoning or cause and effect relationships

oai_token-act-pair · gpt-3.5-turbo

conjunctions and phrases that introduce conditions or reasons

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

•Causality

Because + so,therefore -- seems good

No Comments

Negative Logits

egu

-0.50

erenn

-0.50

abre

-0.49

ãĥ¯ãĥ³

-0.48

Further

-0.47

Basic

-0.47

Eye

-0.45

ãĤ´ãĥ³

-0.44

Versions

-0.44

uty

-0.44

POSITIVE LOGITS

she

0.84

he

0.81

you

0.75

 THEY

0.75

 they

0.74

we

0.73

0.72

SHE

0.72

 thou

0.70

Activations Density 0.663%

phrases indicating reasoning or cause and effect relationships

conjunctions and phrases that introduce conditions or reasons

No Comments

No Known Activations