INDEX

Explanations

keywords related to hidden information or mysteries

oai_token-act-pair · gpt-3.5-turbo

mentions of secrets

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

annis

-0.83

ulf

-0.81

gaard

-0.79

odcast

-0.79

©¶æ

-0.76

baugh

-0.72

lihood

-0.72

odder

-0.71

FG

-0.71

asts

-0.71

POSITIVE LOGITS

ariat

1.32

arial

1.22

 ingredient

0.96

 sauce

0.94

 stash

0.89

 handshake

0.83

 hatch

0.83

rets

0.81

 ballot

0.80

 trove

0.80

Activations Density 0.036%

keywords related to hidden information or mysteries

mentions of secrets

No Comments

No Known Activations