INDEX

Explanations

actionable information or instructions that can be carried out or applied

oai_token-act-pair · gpt-3.5-turbo

phrases indicating capability or potential actions

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

forcing

-0.64

 hating

-0.63

nuts

-0.60

getting

-0.59

gets

-0.58

 Seeking

-0.58

soType

-0.58

Got

-0.56

 Rings

-0.55

bringing

-0.54

POSITIVE LOGITS

be

1.32

't

1.26

 optionally

1.19

isters

1.16

ister

1.08

adian

1.08

NOT

0.99

 accommodate

0.98

 alternatively

0.97

 withstand

0.93

Activations Density 0.109%

actionable information or instructions that can be carried out or applied

phrases indicating capability or potential actions

No Comments

No Known Activations