INDEX

Explanations

the name "Otto"

oai_token-act-pair · gpt-3.5-turbo

references to the individual named "Cotto."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

enez

-1.37

 zones

-0.65

 students

-0.64

 drive

-0.63

 Conver

-0.62

 clos

-0.61

 divergence

-0.61

Zone

-0.60

dep

-0.60

 parallels

-0.59

POSITIVE LOGITS

otto

3.69

otte

2.19

otta

1.48

atto

1.37

oco

1.23

otti

1.19

ott

1.16

ounge

1.06

otine

0.98

omew

0.96

Activations Density 0.010%

the name "Otto"

references to the individual named "Cotto."

No Comments

No Known Activations