INDEX

Explanations

references to people or individuals in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to "people" and their opinions or perceptions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

tnc

-0.85

 srfAttach

-0.83

Accessory

-0.80

uner

-0.70

COMPLE

-0.70

NES

-0.68

efully

-0.68

iary

-0.67

UV

-0.66

Material

-0.65

POSITIVE LOGITS

 smugglers

1.06

who

0.99

 else

0.86

folk

0.86

 perceive

0.84

 wanting

0.81

 underestimate

0.80

 clam

0.79

 realise

0.78

 cared

0.76

Activations Density 0.116%

references to people or individuals in various contexts

references to "people" and their opinions or perceptions

No Comments

No Known Activations