INDEX

Explanations

passive verb phrases

oai_token-act-pair · gpt-3.5-turbo

negative or absent states and contrasts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Ur

-0.78

ONSORED

-0.73

FTWARE

-0.73

EMBER

-0.72

KEN

-0.70

UTF

-0.68

-0.67

ARR

-0.67

IDENT

-0.65

>:

-0.64

POSITIVE LOGITS

 prolifer

1.06

 everywhere

0.86

 routinely

0.84

 disproportionately

0.84

 inherently

0.83

 notoriously

0.82

 typically

0.81

 traditionally

0.80

 generally

0.79

 clustered

0.78

Activations Density 0.622%

passive verb phrases

negative or absent states and contrasts

No Comments

No Known Activations