INDEX

Explanations

the word "cone" and related variations

oai_token-act-pair · gpt-3.5-turbo

references to cones and their related shapes or contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ERAL

-0.72

 GOODMAN

-0.71

estate

-0.67

 Dangerous

-0.67

Ö¼

-0.66

EEK

-0.66

 Berks

-0.66

 Billion

-0.65

 Occupations

-0.64

 Lean

-0.63

POSITIVE LOGITS

 cone

1.47

 cones

1.27

cone

1.05

pole

0.82

otine

0.80

 shaped

0.79

wcs

0.73

lled

0.73

xon

0.72

 disg

0.71

Activations Density 0.010%

the word "cone" and related variations

references to cones and their related shapes or contexts

No Comments

No Known Activations