INDEX

Explanations

instances of potential accusations or attributions of actions by specific entities, such as countries, organizations, or individuals

oai_token-act-pair · gpt-3.5-turbo

phrases that reference actions taken by a specific subject

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

meat

-0.71

edin

-0.71

é¾įåĸļå£«

-0.71

heimer

-0.70

ãĥ´ãĤ¡

-0.70

ãĤ¦ãĤ¹

-0.69

 resil

-0.68

elvet

-0.68

borgh

-0.67

redits

-0.67

POSITIVE LOGITS

 virtue

1.07

 politicians

0.93

products

0.92

 successive

0.89

 policymakers

0.86

 omission

0.86

 individuals

0.84

 whistleblowers

0.83

laws

0.80

 ministers

0.80

Activations Density 0.149%

instances of potential accusations or attributions of actions by specific entities, such as countries, organizations, or individuals

phrases that reference actions taken by a specific subject

No Comments

No Known Activations