INDEX

Explanations

the name "Pence" at various activation levels

oai_token-act-pair · gpt-3.5-turbo

references to political figures, specifically Mike Pence and Tim Kaine

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

ivities

-0.69

actic

-0.69

odon

-0.68

opic

-0.67

variable

-0.65

izations

-0.63

 Flavoring

-0.63

ivation

-0.63

 liner

-0.63

lined

-0.61

POSITIVE LOGITS

 Pence

1.06

mire

0.96

OTUS

0.89

cair

0.78

eln

0.77

lette

0.77

 versa

0.75

etz

0.75

 pard

0.75

tti

0.73

Activations Density 0.009%

the name "Pence" at various activation levels

references to political figures, specifically Mike Pence and Tim Kaine

No Comments

No Known Activations