INDEX

Explanations

words related to errors or problems

oai_token-act-pair · gpt-3.5-turbo

the presence of the suffix "ore" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 partName

-0.66

sid

-0.66

Pwr

-0.64

 registration

-0.63

 certific

-0.63

bun

-0.62

 licences

-0.61

Suc

-0.61

 Bastard

-0.61

 Shooter

-0.60

POSITIVE LOGITS

ore

4.58

ores

2.65

ORE

2.34

oren

1.84

orem

1.83

orean

1.73

ored

1.70

oret

1.59

orer

1.55

orest

1.53

Activations Density 0.005%

words related to errors or problems

the presence of the suffix "ore" in various contexts

No Comments

No Known Activations