INDEX

Explanations

informational phrases prompting action

oai_token-act-pair · gpt-3.5-turbo

requests for additional information

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

 neighb

-0.68

artifacts

-0.63

 lifeless

-0.61

meter

-0.61

opped

-0.61

anmar

-0.60

lihood

-0.60

 impro

-0.58

 throats

-0.56

odied

-0.55

POSITIVE LOGITS

 about

1.18

 regarding

1.15

 ABOUT

1.02

 pertaining

0.98

 concerning

0.96

 About

0.93

About

0.93

about

0.85

 Regarding

0.83

 relating

0.80

Activations Density 0.056%

informational phrases prompting action

requests for additional information

No Comments

No Known Activations