INDEX

Explanations

highly specific terms related to particular subjects or domains

oai_token-act-pair · gpt-3.5-turbo

words related to being dedicated or committed to a specific cause or subject

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Karin

-0.74

ANS

-0.69

FORE

-0.67

 Dunham

-0.66

rett

-0.65

NER

-0.64

Ms

-0.64

 Whitney

-0.64

 Schne

-0.61

 Greenwald

-0.60

POSITIVE LOGITS

icating

0.91

ication

0.88

xual

0.85

icate

0.83

ilation

0.81

rontal

0.80

ications

0.79

aukee

0.79

ombat

0.79

icates

0.77

Activations Density 0.029%

highly specific terms related to particular subjects or domains

words related to being dedicated or committed to a specific cause or subject

No Comments

No Known Activations