INDEX

Explanations

phrases related to giving instructions or commands

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

VERTISEMENT

-0.84

ahime

-0.77

ibrary

-0.76

wcs

-0.76

ighed

-0.76

eco

-0.75

ledged

-0.74

inction

-0.72

iculty

-0.72

eton

-0.71

POSITIVE LOGITS

sir

1.43

 dear

1.26

 gentlemen

1.18

 darling

1.16

 comrade

1.12

 Mister

1.11

 mate

1.09

 buddy

1.05

 please

1.01

huh

0.99

Activations Density 0.149%

phrases related to giving instructions or commands

No Comments

No Known Activations