INDEX

Explanations

names of organizations and institutes

oai_token-act-pair · gpt-3.5-turbo

proper nouns related to organizations, institutions, and reports

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 dearly

-0.70

ntil

-0.67

 unlike

-0.64

 anyways

-0.64

lux

-0.63

 anyway

-0.62

hun

-0.59

ĵĺ

-0.59

 instinct

-0.58

ighting

-0.58

POSITIVE LOGITS

 reveals

1.43

 indicates

1.39

 suggests

1.32

 shows

1.30

 illustrates

1.25

 underscores

1.19

shows

1.18

 confirms

1.17

 demonstrates

1.15

 sheds

1.13

Activations Density 0.258%

names of organizations and institutes

proper nouns related to organizations, institutions, and reports

No Comments

No Known Activations