INDEX

Explanations

mentions of the name "Rom" with varying numerical suffixes

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Rouhani

-0.67

 grounds

-0.66

 dams

-0.64

CPC

-0.64

 attendant

-0.60

 checklist

-0.59

TPS

-0.57

Thumbnail

-0.57

cir

-0.57

 Serious

-0.57

POSITIVE LOGITS

antics

1.38

antically

1.36

ulus

1.29

aji

1.20

ances

1.20

antic

1.20

ijn

1.15

agna

1.13

anes

1.12

ani

1.11

Activations Density 0.035%

mentions of the name "Rom" with varying numerical suffixes

No Comments

No Known Activations